Granola vs Fellow

Agent ops

1. ai-native userAn agent can automatically file each meeting's action items into my project tracker (Linear, Jira, Asana) using the product's API, webhooks, or native automations

   weight 2 · round drawn
   Granola~partialprobed5/10

   Granola offers a documented API/MCP server for programmatic access to meeting notes and a Zapier integration claiming 8,000+ app connections to 'automate what happens after a meeting,' which could route action items to Linear/Jira/Asana, and a native HubSpot sync exists as a precedent for native two-way automation. However, there is no documented native integration or pre-built automation specifically for Linear/Jira/Asana, and MCP/Chat tools shown (search notes, draft email/Slack, calendar events) are read/communication-focused rather than ticket-filing into project trackers. Missing for 10: a documented native Linear/Jira/Asana integration or explicit Zapier recipe/template for action-item ticket creation, and evidence of webhook-triggered ticket creation in these specific trackers.

   • [claimed-docs] “Connect Granola to 8,000+ apps and automate what happens after a meeting”
   • [claimed-docs] “The Granola API lets you build custom integrations with programmatic access to meeting notes, transcripts, and AI summaries.”
   • [claimed-docs] “The Granola API provides programmatic access to your workspace's meeting notes and related data. This RESTful API enables you to integrate G…”
   • [claimed-docs] “Connect your AI tools to your Granola meeting notes with Granola MCP.”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to the official hosted MCP server https://mcp.granola.ai/mcp returned …”
   • [claimed-docs] “Sync Granola meeting notes to HubSpot. Automatically sync notes to CRM records from your meetings using folders.”
   • [claimed-docs] “Ask Chat to: draft and send a follow-up email * draft and send a Slack message * find a time and create a Google Calendar event”

   proof ↗⚑ flag

   Fellow~partialprobed5/10

   Fellow exposes a Developer API (get_action_items, append_note_agenda), webhooks for real-time event notifications, and an MCP server that could be used by an agent to fetch action items and push them elsewhere, giving the building blocks for this automation. However there is no evidence of a native Linear/Jira/Asana integration or documented reference workflow that actually files action items into those specific trackers. Missing for 10: native connectors/integrations to Linear, Jira, or Asana, and a documented example of an agent automatically creating tickets in those tools via the API/webhooks/MCP.

   • [claimed-docs] “List action items with optional filters and pagination.”
   • [claimed-docs] “Webhooks allow you to receive real-time notifications when events occur in Fellow.”
   • [claimed-docs] “Fellow's Model Context Protocol (MCP) Server allows you to ask your AI Assistant about your meetings, without needing to write any code.”
   • [claimed-docs] “pass transcripts to an LLM for tailored recaps and action plans, the API gives you everything you need to make it happen”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to Fellow's hosted MCP server https://fellow.app/mcp returned HTTP 401…”

   proof ↗⚑ flag

2. ai-native userAn official MCP server lets Claude or any MCP client query my meetings — search, transcripts, action items — with OAuth, no glue code

   weight 3 · round drawn
   Granola✓fullprobed9/10

   Granola documents an official hosted MCP server (mcp.granola.ai) supporting search, transcripts, and meeting queries, and a runtime probe confirms the endpoint is live and enforces OAuth via the standard MCP protected-resource flow (401 + WWW-Authenticate pointing to oauth-protected-resource metadata), matching the story's OAuth/no-glue-code requirement. Missing for 10: independent third-party (non-vendor) hands-on confirmation of a successful end-to-end OAuth connection and action-item retrieval.

   • [claimed-docs] “Connect your AI tools to your Granola meeting notes with Granola MCP. Once connected you can use this to: search through your meeting notes”
   • [claimed-docs] “Connect your AI tools to your Granola meeting notes with Granola MCP.”
   • [claimed-docs] “Connect your AI tools to your Granola meeting notes with Granola MCP. Once connected you can use this to: search through your meeting notes,…”
   • [probe] “official MCP server documented at https://docs.granola.ai/help-center/sharing/integrations/mcp”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to the official hosted MCP server https://mcp.granola.ai/mcp returned …”

   proof ↗⚑ flag

   Fellow✓fullprobed9/10

   Fellow ships an official, documented MCP server enabling any MCP client (Claude, ChatGPT, Cursor) to query meetings — search, transcripts, action items — without writing code, and admins control which tools are exposed; a runtime probe confirms the hosted endpoint is live and auth-gated (bearer/OAuth-style), matching the docs. missing for 10: independent third-party review of the OAuth flow specifics beyond the probe's bearer-token confirmation.

   • [claimed-docs] “Ask anything about your meetings - Get answers to questions like "What were my action items this week?"”
   • [claimed-docs] “Fellow’s MCP Connector works with any AI tool that supports connectors, so you're never locked into one platform”
   • [claimed-docs] “As a workspace admin, you decide which MCP tools your members’ AI assistants can use.”
   • [claimed-docs] “Fellow's Model Context Protocol (MCP) Server allows you to ask your AI Assistant about your meetings, without needing to write any code.”
   • [claimed-docs] “Your favorite AI chatbot like Claude, ChatGPT, or Cursor can now tap directly into your meeting recaps, notes, and action items”
   • [claimed-docs] “Ask anything about your meetings - Get answers to questions like "What were my action items this week?" or "What did we decide in last month…”
   • [probe] “official MCP server documented at https://developers.fellow.ai/reference/mcp-server”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to Fellow's hosted MCP server https://fellow.app/mcp returned HTTP 401…”

   proof ↗⚑ flag

3. ai-native userMy scripts can pull transcripts, summaries, and action items from every meeting through a documented API with self-serve keys

   weight 3 · round to Fellow
   Granola~partialprobed7/10

   Granola docs confirm a documented REST API for programmatic access to meeting notes, transcripts, and AI summaries (granola-docs-1,16,33,34), and reference both individual and workspace-level API keys, implying self-serve key generation (granola-docs-28). However, action items are not explicitly called out as a retrievable field, and a probe found no live OpenAPI/swagger spec, suggesting the API reference may be thin or beta (granola-probe-2). missing for 10: explicit documentation of action-item extraction, a public OpenAPI/schema reference, and independent developer corroboration of successful API key self-service and querying.

   • [claimed-docs] “The Granola API provides programmatic access to your workspace's meeting notes and related data.”
   • [claimed-docs] “The Granola API lets you build custom integrations with programmatic access to meeting notes, transcripts, and AI summaries.”
   • [claimed-docs] “The Granola API provides programmatic access to your workspace's meeting notes and related data. This RESTful API enables you to integrate G…”
   • [claimed-docs] “Custom AI agents - Build agents that analyze sentiment across your calls, surface recurring themes in product feedback, or combine meeting c…”
   • [claimed-docs] “Workspace admins on Business and Enterprise plans can also create workspace API keys. These belong to the workspace rather than an individua…”
   • [probe] “PROBE openapi: all candidate paths 404 (https://docs.granola.ai/openapi.json, https://docs.granola.ai/swagger.json, https://docs.granola.ai/…”

   proof ↗⚑ flag

   Fellow✓fullprobed8/10

   Fellow documents a REST Developer API with self-serve API key generation (docs-2,15,16) and endpoints for recordings/transcripts, action items, and notes (docs-4,5,6,31), plus webhooks for real-time events. Runtime probe confirms the API/MCP endpoints are live and properly authenticated, not vaporware. missing for 10: no public OpenAPI/swagger spec found (404s), and no independent third-party developer report of building against the API.

   • [claimed-docs] “The Developer API opens that data to you: transcripts, structured notes, and more, all available through standard REST calls.”
   • [claimed-docs] “All API requests must include a valid API key in the Authorization header.”
   • [claimed-docs] “List recordings with optional filters and pagination.”
   • [claimed-docs] “List action items with optional filters and pagination.”
   • [claimed-docs] “Include your API key in the `X-API-KEY` header of every request”
   • [claimed-docs] “Generate a new API key for your application, giving it a descriptive name.”
   • [claimed-docs] “Webhooks allow you to receive real-time notifications when events occur in Fellow.”
   • [claimed-docs] “pass transcripts to an LLM for tailored recaps and action plans, the API gives you everything you need to make it happen”
   • [probe] “PROBE llms.txt: HTTP 200 at https://developers.fellow.ai/llms.txt # Fellow Documentation > Fellow’s API provides access to meeting notes an…”
   • [probe] “PROBE openapi: all candidate paths 404 (https://developers.fellow.ai/openapi.json, https://developers.fellow.ai/swagger.json, https://develo…”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to Fellow's hosted MCP server https://fellow.app/mcp returned HTTP 401…”

   proof ↗⚑ flag

Agent access

1. ai-native userPoint an agent at llms.txt or agent-oriented docs

   weight 2 · round to Granola
   Granola✓fullprobed9/10

   A live llms.txt is confirmed via direct HTTP probe (200 OK) at docs.granola.ai/llms.txt, and Granola also documents agent-oriented resources (API for custom agents, an official OAuth-protected MCP server verified live via runtime probe), giving agents multiple discoverable, machine-readable entry points into Granola's docs and data. Missing for 10: no independent/community confirmation of an agent actually consuming llms.txt successfully in practice.

   • [probe] “PROBE llms.txt: HTTP 200 at https://docs.granola.ai/llms.txt # Granola Docs & Help Center - [Granola Help Center](https://docs.granola.ai/h…”
   • [probe] “official MCP server documented at https://docs.granola.ai/help-center/sharing/integrations/mcp”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to the official hosted MCP server https://mcp.granola.ai/mcp returned …”
   • [claimed-docs] “The Granola API lets you build custom integrations with programmatic access to meeting notes, transcripts, and AI summaries.”
   • [claimed-docs] “Custom AI agents - Build agents that analyze sentiment across your calls, surface recurring themes in product feedback, or combine meeting c…”

   proof ↗⚑ flag

   Fellow✓fullprobed8/10

   A live probe confirms Fellow serves a proper llms.txt file at https://developers.fellow.ai/llms.txt (HTTP 200) summarizing its API for agent consumption, and the broader developer docs (API reference, MCP server docs) are structured for both human and agent use. Missing for 10: no evidence of a companion llms-full.txt or independent third-party confirmation that agents successfully consume this file in practice.

   • [probe] “PROBE llms.txt: HTTP 200 at https://developers.fellow.ai/llms.txt # Fellow Documentation > Fellow’s API provides access to meeting notes an…”
   • [claimed-docs] “pass transcripts to an LLM for tailored recaps and action plans, the API gives you everything you need to make it happen”
   • [probe] “official MCP server documented at https://developers.fellow.ai/reference/mcp-server”

   proof ↗⚑ flag

2. ai-native userRun the product headlessly / in CI for automation

   weight 2 · round to Granola
   Granola~partialprobed3/10

   Granola exposes a REST API and an MCP server for programmatic access to meeting notes/transcripts, which could be scripted into automated pipelines, but the core product is a desktop app that captures live meeting audio and there is no documented CLI, headless mode, or CI-specific workflow. Missing for 10: explicit headless/CI execution mode, CLI tooling, and any example of running Granola non-interactively in an automation pipeline.

   • [claimed-docs] “The Granola API provides programmatic access to your workspace's meeting notes and related data.”
   • [claimed-docs] “The Granola API lets you build custom integrations with programmatic access to meeting notes, transcripts, and AI summaries.”
   • [claimed-docs] “The Granola API provides programmatic access to your workspace's meeting notes and related data. This RESTful API enables you to integrate G…”
   • [probe] “official MCP server documented at https://docs.granola.ai/help-center/sharing/integrations/mcp”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to the official hosted MCP server https://mcp.granola.ai/mcp returned …”

   proof ↗⚑ flag

   Fellow—none0/10

   The axis applies to this product kind (peer products hold positive or none verdicts on this story), so lack of evidence for an applicable capability is "none", never "na". (na/none harmonized at arena bring-up — see pipeline/scripts/na-harmonize.ts.)

   ⚑ flag

3. ai-native userPlug MCP servers into this product so it can use their tools

   weight 3 · round drawn
   Granola—none0/10

   All evidence describes Granola exposing its own MCP server (and API) so other AI tools can pull Granola's meeting data — the reverse of the story, which asks whether Granola itself can consume external MCP servers as a client to gain new tools. No evidence shows Granola's Chat or any other feature accepting/plugging in third-party MCP servers.

   • [claimed-docs] “Connect your AI tools to your Granola meeting notes with Granola MCP. Once connected you can use this to: search through your meeting notes”
   • [claimed-docs] “Connect your AI tools to your Granola meeting notes with Granola MCP.”
   • [claimed-docs] “Connect your AI tools to your Granola meeting notes with Granola MCP. Once connected you can use this to: search through your meeting notes,…”
   • [probe] “official MCP server documented at https://docs.granola.ai/help-center/sharing/integrations/mcp”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to the official hosted MCP server https://mcp.granola.ai/mcp returned …”

   proof ↗⚑ flag

   Fellow—none0/10

   All evidence describes Fellow acting as an MCP *server*, exposing its own meeting data/tools to external AI assistants (Claude, ChatGPT, Cursor) — the opposite direction of this story, which asks whether a user can plug external MCP servers into Fellow so Fellow itself can use their tools. No evidence shows Fellow consuming or connecting to third-party MCP servers.

   ⚑ flag

4. ai-native userConnect an agent via an official MCP server

   weight 3 · round to Fellow
   Granola✓fullprobed8/10

   Granola documents an official MCP server ('Granola MCP') for connecting AI tools/agents to meeting notes (search, list_meetings, get_meeting_transcript, query_granola_meetings), and a runtime probe confirms the hosted endpoint https://mcp.granola.ai/mcp is live and correctly speaks MCP's OAuth-protected resource flow. Missing for 10: independent/community hands-on confirmation of a successful agent connection beyond the vendor docs and probe.

   • [claimed-docs] “Connect your AI tools to your Granola meeting notes with Granola MCP. Once connected you can use this to: search through your meeting notes”
   • [claimed-docs] “Connect your AI tools to your Granola meeting notes with Granola MCP.”
   • [claimed-docs] “Connect your AI tools to your Granola meeting notes with Granola MCP. Once connected you can use this to: search through your meeting notes,…”
   • [probe] “official MCP server documented at https://docs.granola.ai/help-center/sharing/integrations/mcp”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to the official hosted MCP server https://mcp.granola.ai/mcp returned …”

   proof ↗⚑ flag

   Fellow✓fullprobed9/10

   Fellow ships an official, documented MCP Server (developers.fellow.ai/reference/mcp-server) that lets AI assistants like Claude, ChatGPT, or Cursor query meeting transcripts and action items without code, with admin-level tool controls; a runtime probe confirms the hosted endpoint (fellow.app/mcp) is live and properly bearer-gated as documented. missing for 10: independent third-party/community usage reports beyond the vendor docs and the single runtime probe.

   • [claimed-docs] “Fellow's Model Context Protocol (MCP) Server allows you to ask your AI Assistant about your meetings, without needing to write any code.”
   • [claimed-docs] “Fellow’s MCP Connector works with any AI tool that supports connectors, so you're never locked into one platform”
   • [claimed-docs] “As a workspace admin, you decide which MCP tools your members’ AI assistants can use.”
   • [claimed-docs] “Your favorite AI chatbot like Claude, ChatGPT, or Cursor can now tap directly into your meeting recaps, notes, and action items”
   • [probe] “official MCP server documented at https://developers.fellow.ai/reference/mcp-server”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to Fellow's hosted MCP server https://fellow.app/mcp returned HTTP 401…”

   proof ↗⚑ flag

5. ai-native userDrive the product through a documented public API

   weight 3 · round to Fellow
   Granola✓fullprobed7/10

   Granola documents a RESTful public API for programmatic access to meeting notes, transcripts, and summaries, including workspace-level API keys for admins, and a live official MCP server (confirmed via runtime probe) that supports OAuth-based agent access to meeting data. This is solid documented evidence of API-driven, agentic use of the product beyond the base app UI. Missing for 10: a formal OpenAPI/reference spec (probe found 404s for openapi.json/swagger.json) and independent third-party developer reports validating real-world API integration.

   • [claimed-docs] “The Granola API provides programmatic access to your workspace's meeting notes and related data. This RESTful API enables you to integrate G…”
   • [claimed-docs] “Workspace admins on Business and Enterprise plans can also create workspace API keys. These belong to the workspace rather than an individua…”
   • [claimed-docs] “Custom AI agents - Build agents that analyze sentiment across your calls, surface recurring themes in product feedback, or combine meeting c…”
   • [probe] “official MCP server documented at https://docs.granola.ai/help-center/sharing/integrations/mcp”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to the official hosted MCP server https://mcp.granola.ai/mcp returned …”
   • [probe] “PROBE openapi: all candidate paths 404 (https://docs.granola.ai/openapi.json, https://docs.granola.ai/swagger.json, https://docs.granola.ai/…”

   proof ↗⚑ flag

   Fellow✓fullprobed8/10

   Fellow ships a documented Developer API with REST endpoints (recordings, action items, note editing), API-key authentication, and webhooks, giving AI-native users a clear path to programmatic control and even LLM-facing MCP integration. missing for 10: a public OpenAPI/swagger spec (probes returned 404) and independent/third-party developer corroboration beyond vendor docs.

   • [claimed-docs] “The Developer API opens that data to you: transcripts, structured notes, and more, all available through standard REST calls.”
   • [claimed-docs] “All API requests must include a valid API key in the Authorization header.”
   • [claimed-docs] “List recordings with optional filters and pagination.”
   • [claimed-docs] “List action items with optional filters and pagination.”
   • [claimed-docs] “Add Fellow Markdown to the end of a note's agenda, leaving the rest of the content in place.”
   • [claimed-docs] “Webhooks allow you to receive real-time notifications when events occur in Fellow.”
   • [probe] “PROBE openapi: all candidate paths 404 (https://developers.fellow.ai/openapi.json, https://developers.fellow.ai/swagger.json, https://develo…”

   proof ↗⚑ flag

6. ai-native userIssue scoped/least-privilege API credentials for an agent

   weight 2 · round to Fellow
   Granola—none0/10

   Granola documents API keys (personal and workspace-level) and an OAuth-based MCP server, but there is no evidence of scoped or least-privilege credential issuance — e.g., no read-only vs write scopes, no per-agent permission granularity, and workspace keys are described as broad and non-expiring rather than restrictable.

   • [claimed-docs] “Workspace admins on Business and Enterprise plans can also create workspace API keys. These belong to the workspace rather than an individua…”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to the official hosted MCP server https://mcp.granola.ai/mcp returned …”
   • [probe] “official MCP server documented at https://docs.granola.ai/help-center/sharing/integrations/mcp”

   proof ↗⚑ flag

   Fellow~partialclaimed5/10

   Fellow supports API key auth and lets workspace admins control which MCP tools an AI assistant can access (fellow-docs-10), which is a coarse form of least-privilege control for agents. However there's no documented granular scoping (e.g., read-only vs write, resource-level permissions, expiry) for individual API keys — Super Admin keys in fact grant broad on-behalf-of access across the workspace, the opposite of least privilege. Missing for 10: explicit API key scope/permission levels, token expiration/revocation controls, and fine-grained per-resource credential issuance for agents.

   • [claimed-docs] “As a workspace admin, you decide which MCP tools your members’ AI assistants can use.”
   • [claimed-docs] “Super Admin API keys can make requests on behalf of any user in the workspace using the `X-On-Behalf-Of` header.”
   • [claimed-docs] “Fellow’s Super Admin API lets designated Enterprise‑plan administrators retrieve, export, and delete data across the entire workspace”
   • [claimed-docs] “Generate a new API key for your application, giving it a descriptive name.”
   • [claimed-docs] “All API requests must include a valid API key in the Authorization header.”

   proof ↗⚑ flag

7. ai-native userBuild against official SDKs

   weight 2 · round to Granola
   Granola~partialprobed5/10

   Granola exposes a documented RESTful API and an official hosted MCP server (confirmed live via runtime probe with OAuth), which lets AI-native users build custom agents/integrations, but there is no evidence of official language SDKs, client libraries, or a discoverable OpenAPI/swagger spec (probe found only 404s). Missing for 10: published SDKs in common languages, an OpenAPI spec, and independent developer corroboration of building against the API.

   • [claimed-docs] “The Granola API lets you build custom integrations with programmatic access to meeting notes, transcripts, and AI summaries.”
   • [claimed-docs] “The Granola API provides programmatic access to your workspace's meeting notes and related data. This RESTful API enables you to integrate G…”
   • [claimed-docs] “Custom AI agents - Build agents that analyze sentiment across your calls, surface recurring themes in product feedback, or combine meeting c…”
   • [claimed-docs] “Connect your AI tools to your Granola meeting notes with Granola MCP. Once connected you can use this to: search through your meeting notes”
   • [probe] “official MCP server documented at https://docs.granola.ai/help-center/sharing/integrations/mcp”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to the official hosted MCP server https://mcp.granola.ai/mcp returned …”
   • [probe] “PROBE openapi: all candidate paths 404 (https://docs.granola.ai/openapi.json, https://docs.granola.ai/swagger.json, https://docs.granola.ai/…”

   proof ↗⚑ flag

   Fellow—none0/10

   Fellow documents a REST Developer API with API-key auth, webhooks, and endpoints for recordings/notes/action items, but no evidence names an official SDK (e.g., Python/JS client library); the openapi.json/swagger probe returned 404 across all standard paths, suggesting no formal spec-based SDK generation either.

   • [claimed-docs] “The Developer API opens that data to you: transcripts, structured notes, and more, all available through standard REST calls.”
   • [claimed-docs] “All API requests must include a valid API key in the Authorization header.”
   • [claimed-docs] “Include your API key in the `X-API-KEY` header of every request”
   • [claimed-docs] “Generate a new API key for your application, giving it a descriptive name.”
   • [probe] “PROBE openapi: all candidate paths 404 (https://developers.fellow.ai/openapi.json, https://developers.fellow.ai/swagger.json, https://develo…”

   proof ↗⚑ flag

8. ai-native userSubscribe to events via webhooks

   weight 2 · round to Fellow
   Granola—none0/10

   Evidence covers a REST API, MCP server, Zapier/HubSpot integrations, and chat workflows, but nothing in the pack describes any webhook subscription mechanism for event notifications. Zapier could theoretically enable trigger-based automation, but no documentation confirms Granola exposes outbound webhooks for developers to subscribe to events directly. missing for 10: documented webhook endpoints, event types, subscription/registration mechanism, payload schema.

   • [claimed-docs] “The Granola API lets you build custom integrations with programmatic access to meeting notes, transcripts, and AI summaries.”
   • [claimed-docs] “Connect Granola to 8,000+ apps and automate what happens after a meeting”
   • [probe] “PROBE openapi: all candidate paths 404 (https://docs.granola.ai/openapi.json, https://docs.granola.ai/swagger.json, https://docs.granola.ai/…”

   proof ↗⚑ flag

   Fellow✓fullclaimed7/10

   Fellow's Developer API explicitly documents webhooks that deliver real-time HTTP POST notifications when events occur, which directly satisfies the subscribe-to-events story. Missing for 10: a list of supported event types/payload schema, and independent/hands-on confirmation of webhook delivery reliability.

   • [claimed-docs] “Fellow sends an HTTP POST request to the URL you specify with details about the event.”
   • [claimed-docs] “Webhooks allow you to receive real-time notifications when events occur in Fellow.”

   proof ↗⚑ flag

Agentic features

1. ai-native userGet AI-generated insights and suggestions from my data inside the product

   weight 2 · round drawn
   Granola✓fullclaimed8/10

   Granola Chat lets users ask questions and get instant AI-generated insights, spot patterns across calls, and generate follow-up outputs directly inside the product, with Recipes providing reusable insight-generating prompt templates. This is native, first-party AI-native functionality well documented across multiple docs pages. Missing for 10: independent/hands-on validation of insight quality and accuracy beyond vendor docs.

   • [claimed-docs] “Chat with your meeting notes to recall meeting details, spot patterns across customer calls, and generate follow-up outputs.”
   • [claimed-docs] “Granola Chat can help you complete the work that comes after a meeting... draft and send a follow-up email”
   • [claimed-docs] “Recipes are reusable prompt templates that help you quickly get insights from your meeting notes without typing the same questions repeatedl…”
   • [claimed-docs] “Granola Chat lets you ask questions about your meetings and get instant insights from your notes or transcripts.”
   • [claimed-docs] “draft and send a follow-up email" ... "draft and send a Slack message" ... "find a time and create a Google Calendar event”
   • [claimed-docs] “Ask Chat to: draft and send a follow-up email * draft and send a Slack message * find a time and create a Google Calendar event”
   • [claimed-docs] “Granola Chat can help you complete the work that comes after a meeting, without switching between your notes and other apps.”

   proof ↗⚑ flag

   Fellow✓fullprobed8/10

   Fellow auto-generates AI meeting summaries, action items, decisions, and topic-based minutes for every meeting, plus MCP server and API access to let AI assistants query and reason over that data ('What were my action items this week?'). Runtime probe confirms the MCP server is live and functioning as documented (bearer-gated, not broken). missing for 10: independent third-party review of insight quality/accuracy, and no in-product analytics/dashboard example beyond summaries and action items.

   • [claimed-docs] “You automatically receive a clear meeting summary, suggested AI action items, decisions, and topic-based minutes for every meeting.”
   • [claimed-docs] “Ask anything about your meetings - Get answers to questions like "What were my action items this week?"”
   • [claimed-docs] “Ask anything about your meetings - Get answers to questions like "What were my action items this week?" or "What did we decide in last month…”
   • [claimed-docs] “Your favorite AI chatbot like Claude, ChatGPT, or Cursor can now tap directly into your meeting recaps, notes, and action items”
   • [claimed-docs] “pass transcripts to an LLM for tailored recaps and action plans, the API gives you everything you need to make it happen”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to Fellow's hosted MCP server https://fellow.app/mcp returned HTTP 401…”

   proof ↗⚑ flag

2. ai-native userSet up automations that run autonomously in the background

   weight 2 · round drawn
   Granola~partialprobed5/10

   Granola supports automations that fire after meetings via Zapier (8,000+ app connections) and HubSpot auto-sync, and offers an API/MCP server for building custom 'agents' that analyze calls—these run without manual intervention once configured. However, there's no evidence of a native scheduler or persistent background agent loop inside Granola itself; the 'autonomy' comes from third-party automation platforms (Zapier/HubSpot) or requires the user to build their own agent via API/MCP rather than Granola running one natively. Missing for 10: native in-app automation/scheduling engine, evidence of continuously running background agents (not just event-triggered post-meeting syncs), and independent confirmation these automations run reliably unattended.

   • [claimed-docs] “Connect Granola to 8,000+ apps and automate what happens after a meeting”
   • [claimed-docs] “Sync Granola meeting notes to HubSpot. Automatically sync notes to CRM records from your meetings using folders.”
   • [claimed-docs] “You can set up Granola to automatically save any notes that are added to Granola folders to the relevant contacts, companies, or deals on Hu…”
   • [claimed-docs] “Custom AI agents - Build agents that analyze sentiment across your calls, surface recurring themes in product feedback, or combine meeting c…”
   • [claimed-docs] “The Granola API lets you build custom integrations with programmatic access to meeting notes, transcripts, and AI summaries.”
   • [probe] “official MCP server documented at https://docs.granola.ai/help-center/sharing/integrations/mcp”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to the official hosted MCP server https://mcp.granola.ai/mcp returned …”

   proof ↗⚑ flag

   Fellow~partialclaimed5/10

   Fellow automatically generates summaries, action items, and decisions in the background after every meeting without manual triggering (fellow-docs-23), and its webhook system plus Developer API let external automations fire in real time when events occur (fellow-docs-3, fellow-docs-17, fellow-docs-31). However, there's no evidence of a native automation/workflow builder, scheduled triggers, or rule-based autonomous actions within Fellow itself — automation would need to be built externally using the webhook/API primitives. Missing for 10: a first-party automation/rules engine, scheduling capability, and hands-on evidence of end-to-end autonomous workflows beyond passive notification/data-access APIs.

   • [claimed-docs] “You automatically receive a clear meeting summary, suggested AI action items, decisions, and topic-based minutes for every meeting.”
   • [claimed-docs] “Fellow sends an HTTP POST request to the URL you specify with details about the event.”
   • [claimed-docs] “Webhooks allow you to receive real-time notifications when events occur in Fellow.”
   • [claimed-docs] “pass transcripts to an LLM for tailored recaps and action plans, the API gives you everything you need to make it happen”
   • [claimed-docs] “The Developer API opens that data to you: transcripts, structured notes, and more, all available through standard REST calls.”

   proof ↗⚑ flag

3. ai-native userDelegate tasks to a built-in AI assistant inside the product

   weight 3 · round to Granola
   Granola✓fullclaimed7/10

   Granola ships a built-in 'Granola Chat' assistant that users can delegate post-meeting tasks to, e.g. drafting and sending follow-up emails/Slack messages and creating calendar events directly from meeting notes, plus reusable 'Recipes' prompt templates for repeated tasks. This is documented across multiple first-party doc pages describing concrete agentic actions triggered from chat. Missing for 10: independent/hands-on verification that these delegated actions (sending email, creating calendar events) actually execute reliably, and more detail on assistant autonomy/error-handling beyond vendor docs.

   • [claimed-docs] “Chat with your meeting notes to recall meeting details, spot patterns across customer calls, and generate follow-up outputs.”
   • [claimed-docs] “Granola Chat can help you complete the work that comes after a meeting... draft and send a follow-up email”
   • [claimed-docs] “Granola Chat lets you ask questions about your meetings and get instant insights from your notes or transcripts.”
   • [claimed-docs] “draft and send a follow-up email" ... "draft and send a Slack message" ... "find a time and create a Google Calendar event”
   • [claimed-docs] “Ask Chat to: draft and send a follow-up email * draft and send a Slack message * find a time and create a Google Calendar event”
   • [claimed-docs] “Granola Chat can help you complete the work that comes after a meeting, without switching between your notes and other apps.”
   • [claimed-docs] “Recipes are reusable prompt templates that help you quickly get insights from your meeting notes without typing the same questions repeatedl…”

   proof ↗⚑ flag

   Fellow~partialclaimed4/10

   Fellow's AI automatically generates meeting summaries, action items, and decisions (fellow-docs-23, fellow-docs-31), which resembles delegated AI task execution, but there is no evidence of a conversational 'built-in assistant' inside the product that a user can direct with arbitrary tasks — the MCP server (fellow-docs-18, fellow-docs-26) instead lets *external* assistants like Claude/ChatGPT query Fellow's data, which is the inverse of a built-in assistant. Missing for 10: an in-app conversational assistant UI, evidence of multi-step task delegation beyond automatic note/action-item generation, and confirmation this assistant runs natively rather than via third-party AI tools.

   • [claimed-docs] “You automatically receive a clear meeting summary, suggested AI action items, decisions, and topic-based minutes for every meeting.”
   • [claimed-docs] “pass transcripts to an LLM for tailored recaps and action plans, the API gives you everything you need to make it happen”
   • [claimed-docs] “Fellow's Model Context Protocol (MCP) Server allows you to ask your AI Assistant about your meetings, without needing to write any code.”
   • [claimed-docs] “Your favorite AI chatbot like Claude, ChatGPT, or Cursor can now tap directly into your meeting recaps, notes, and action items”
   • [claimed-docs] “Ask anything about your meetings - Get answers to questions like "What were my action items this week?" or "What did we decide in last month…”

   proof ↗⚑ flag

4. ai-native userOperate the product with natural-language commands

   weight 2 · round to Granola
   Granola✓fullclaimed7/10

   Granola Chat lets users interact via natural-language queries to recall meeting details, spot patterns, and trigger actions like drafting/sending a follow-up email, drafting a Slack message, or creating a calendar event—clear NL-command operation within the product. missing for 10: independent/hands-on corroboration of chat-driven command execution, and detail on reliability/limits of the NL command surface beyond first-party docs.

   • [claimed-docs] “Chat with your meeting notes to recall meeting details, spot patterns across customer calls, and generate follow-up outputs.”
   • [claimed-docs] “Granola Chat can help you complete the work that comes after a meeting... draft and send a follow-up email”
   • [claimed-docs] “Granola Chat lets you ask questions about your meetings and get instant insights from your notes or transcripts.”
   • [claimed-docs] “draft and send a follow-up email" ... "draft and send a Slack message" ... "find a time and create a Google Calendar event”
   • [claimed-docs] “Ask Chat to: draft and send a follow-up email * draft and send a Slack message * find a time and create a Google Calendar event”
   • [claimed-docs] “Granola Chat can help you complete the work that comes after a meeting, without switching between your notes and other apps.”

   proof ↗⚑ flag

   Fellow~partialprobed7/10

   Fellow ships an official, documented MCP server that works with AI assistants (Claude, ChatGPT, Cursor) to let users ask natural-language questions like "What were my action items this week?", and a runtime probe confirms the hosted MCP endpoint is live and bearer-gated as documented. However, the evidence shows this is primarily a query/read interface (search meetings, get action items, get transcripts) rather than full natural-language control over write/agentic actions. Missing for 10: evidence of NL-driven write actions (e.g., scheduling, editing notes, assigning action items) executed through natural-language commands rather than just Q&A retrieval.

   • [claimed-docs] “Ask anything about your meetings - Get answers to questions like "What were my action items this week?"”
   • [claimed-docs] “Fellow’s MCP Connector works with any AI tool that supports connectors, so you're never locked into one platform”
   • [claimed-docs] “Fellow's Model Context Protocol (MCP) Server allows you to ask your AI Assistant about your meetings, without needing to write any code.”
   • [claimed-docs] “Your favorite AI chatbot like Claude, ChatGPT, or Cursor can now tap directly into your meeting recaps, notes, and action items”
   • [claimed-docs] “Ask anything about your meetings - Get answers to questions like "What were my action items this week?" or "What did we decide in last month…”
   • [probe] “official MCP server documented at https://developers.fellow.ai/reference/mcp-server”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to Fellow's hosted MCP server https://fellow.app/mcp returned HTTP 401…”

   proof ↗⚑ flag

Api quality

1. ai-native userExplore an interactive API reference with runnable examples

   weight 2 · round to Fellow
   Granola—none0/10

   There's a mention of a RESTful API and MCP server, but no evidence of an interactive API reference with runnable examples; probes explicitly show no OpenAPI/swagger spec is exposed (all candidate paths 404), and no docs mention interactive playgrounds or code samples runnable in-browser.

   • [probe] “PROBE openapi: all candidate paths 404 (https://docs.granola.ai/openapi.json, https://docs.granola.ai/swagger.json, https://docs.granola.ai/…”
   • [claimed-docs] “The Granola API provides programmatic access to your workspace's meeting notes and related data. This RESTful API enables you to integrate G…”
   • [claimed-docs] “The Granola API lets you build custom integrations with programmatic access to meeting notes, transcripts, and AI summaries.”

   proof ↗⚑ flag

   Fellow~partialprobed3/10

   Fellow publishes a structured Developer API reference with per-endpoint pages (recordings, action items, note editing, webhooks) suggesting a ReadMe.io-style docs site, but there is no confirmed evidence of a live 'try it' interactive console or runnable code examples — the openapi.json/swagger.json probe returned 404 across all candidate paths, indicating no public machine-readable spec to power an interactive explorer. missing for 10: explicit runnable 'try it' console evidence, public OpenAPI/Swagger spec, and independent confirmation of interactivity.

   • [claimed-docs] “List recordings with optional filters and pagination.”
   • [claimed-docs] “List action items with optional filters and pagination.”
   • [claimed-docs] “Add Fellow Markdown to the end of a note's agenda, leaving the rest of the content in place.”
   • [claimed-docs] “The Developer API opens that data to you: transcripts, structured notes, and more, all available through standard REST calls.”
   • [probe] “PROBE openapi: all candidate paths 404 (https://developers.fellow.ai/openapi.json, https://developers.fellow.ai/swagger.json, https://develo…”

   proof ↗⚑ flag

2. ai-native userDownload a machine-readable API spec (OpenAPI or equivalent)

   weight 2 · round drawn
   Granola—none0/10

   Granola documents a REST API and MCP integration but explicit probes show no OpenAPI/Swagger spec is published at any standard location (all candidate paths returned 404), and no machine-readable API spec is referenced anywhere in the docs.

   • [probe] “PROBE openapi: all candidate paths 404 (https://docs.granola.ai/openapi.json, https://docs.granola.ai/swagger.json, https://docs.granola.ai/…”
   • [claimed-docs] “The Granola API provides programmatic access to your workspace's meeting notes and related data. This RESTful API enables you to integrate G…”

   proof ↗⚑ flag

   Fellow—none0/10

   Fellow's developer docs use ReadMe-style reference pages but the probe explicitly found no OpenAPI/Swagger spec at any standard location (all 404), and no evidence of a downloadable machine-readable spec is cited elsewhere.

   • [probe] “PROBE openapi: all candidate paths 404 (https://developers.fellow.ai/openapi.json, https://developers.fellow.ai/swagger.json, https://develo…”

   proof ↗⚑ flag

3. ai-native userRely on versioned APIs with a documented deprecation policy

   weight 2 · round drawn
   Granola—none0/10

   The evidence shows a Granola REST API and MCP server exist, but there is no documentation of API versioning scheme or a deprecation policy; an openapi.json probe returned 404s, suggesting no formal API spec/versioning surface is published. missing for 10: versioning scheme documentation, deprecation policy/notice process, changelog or migration guides.

   • [claimed-docs] “The Granola API provides programmatic access to your workspace's meeting notes and related data. This RESTful API enables you to integrate G…”
   • [probe] “PROBE openapi: all candidate paths 404 (https://docs.granola.ai/openapi.json, https://docs.granola.ai/swagger.json, https://docs.granola.ai/…”
   • [claimed-docs] “Workspace admins on Business and Enterprise plans can also create workspace API keys. These belong to the workspace rather than an individua…”

   proof ↗⚑ flag

   Fellow—none0/10

   Evidence documents API endpoints, auth, and webhooks, but there is no mention anywhere of API versioning scheme or a documented deprecation policy; an openapi spec probe even 404'd, suggesting no formal versioned spec is published.

   • [probe] “PROBE openapi: all candidate paths 404 (https://developers.fellow.ai/openapi.json, https://developers.fellow.ai/swagger.json, https://develo…”
   • [claimed-docs] “The Developer API opens that data to you: transcripts, structured notes, and more, all available through standard REST calls.”
   • [claimed-docs] “All API requests must include a valid API key in the Authorization header.”

   proof ↗⚑ flag

1. ai-native userPerform bulk operations across many items at once

   weight 2 · round to Fellow
   Granola~partialprobed4/10

   Granola offers bulk *export* of all notes to CSV and API/MCP access that can read across the whole workspace (search, list_meetings), which supports bulk analysis, but there's no documented capability for bulk mutating operations (e.g., batch delete, batch re-tag, batch share) across many items at once. missing for 10: explicit bulk write/update operations, batch API endpoints for multi-item actions, evidence of an agent actually performing such an operation.

   • [claimed-docs] “Export your Granola meeting notes to CSV. Get note titles, summaries, transcripts, and details.”
   • [claimed-docs] “You're able to export historical data from Granola by going to Settings, clicking Profile in the sidebar (under Preferences), and clicking G…”
   • [claimed-docs] “The Granola API provides programmatic access to your workspace's meeting notes and related data. This RESTful API enables you to integrate G…”
   • [probe] “official MCP server documented at https://docs.granola.ai/help-center/sharing/integrations/mcp”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to the official hosted MCP server https://mcp.granola.ai/mcp returned …”

   proof ↗⚑ flag

   Fellow~partialclaimed5/10

   Fellow's Developer API supports listing recordings/action items with filters and pagination (bulk read), and the Super Admin API explicitly allows retrieving, exporting, and deleting data across the entire workspace, which covers workspace-wide bulk read/delete for admins. However, there's no documented endpoint for bulk create/update/write operations (e.g., batch-editing many action items or notes in one call) — missing for 10: bulk write/update endpoints, batch action-item creation, evidence of true multi-item mutation in a single API call.

   • [claimed-docs] “List recordings with optional filters and pagination.”
   • [claimed-docs] “List action items with optional filters and pagination.”
   • [claimed-docs] “Fellow’s Super Admin API lets designated Enterprise‑plan administrators retrieve, export, and delete data across the entire workspace”
   • [claimed-docs] “Build admin dashboards that display data from a specific user's perspective”

   proof ↗⚑ flag

2. ai-native userDefine rules that trigger actions automatically on events

   weight 3 · round drawn
   Granola~partialclaimed4/10

   Granola supports event-triggered automation only via third-party integrations: Zapier (8,000+ apps, 'automate what happens after a meeting') and HubSpot folder-based auto-sync ('automatically save any notes added to Granola folders to relevant contacts/deals'). These act as rule-like triggers but are not a native in-product rules/automation engine, and Chat 'workflows' are manually invoked rather than event-triggered. Missing for 10: a native rules/automation builder within Granola itself, documentation of custom trigger-condition-action definitions, and evidence of broader event types beyond folder-add/meeting-end.

   • [claimed-docs] “Connect Granola to 8,000+ apps and automate what happens after a meeting”
   • [claimed-docs] “You can set up Granola to automatically save any notes that are added to Granola folders to the relevant contacts, companies, or deals on Hu…”
   • [claimed-docs] “Automatically sync notes to CRM records from your meetings using folders.”
   • [claimed-docs] “draft and send a follow-up email" ... "draft and send a Slack message" ... "find a time and create a Google Calendar event”

   proof ↗⚑ flag

   Fellow~partialclaimed4/10

   Fellow offers webhooks that fire real-time notifications on events, which can be used to build external automations, but there is no evidence of a native rules/trigger builder within Fellow itself that lets a user define 'if event X then action Y' logic. Missing for 10: a documented in-product automation/rules engine, examples of user-configurable triggers-to-actions, and any built-in action execution beyond webhook delivery.

   • [claimed-docs] “Fellow sends an HTTP POST request to the URL you specify with details about the event.”
   • [claimed-docs] “Webhooks allow you to receive real-time notifications when events occur in Fellow.”
   • [claimed-docs] “pass transcripts to an LLM for tailored recaps and action plans, the API gives you everything you need to make it happen”

   proof ↗⚑ flag

3. ai-native userSchedule recurring jobs or workflows

   weight 2 · round drawn
   Granola—none0/10

   Evidence shows Granola offers APIs, MCP, Zapier/HubSpot integrations, and one-off Chat workflows (draft emails, Slack messages, calendar events), but nothing describes scheduling recurring jobs or automated workflows to run on a cadence within Granola itself — recurring automation would rely entirely on third-party tools like Zapier, not a native Granola feature.

   • [claimed-docs] “Connect Granola to 8,000+ apps and automate what happens after a meeting”
   • [claimed-docs] “draft and send a follow-up email" ... "draft and send a Slack message" ... "find a time and create a Google Calendar event”
   • [claimed-docs] “Ask Chat to: draft and send a follow-up email * draft and send a Slack message * find a time and create a Google Calendar event”
   • [claimed-docs] “The Granola API lets you build custom integrations with programmatic access to meeting notes, transcripts, and AI summaries.”
   • [claimed-docs] “Custom AI agents - Build agents that analyze sentiment across your calls, surface recurring themes in product feedback, or combine meeting c…”

   proof ↗⚑ flag

   Fellow—none0/10

   Fellow is a meeting-notes/AI assistant product with a REST API, webhooks, and MCP server, but nothing in the evidence describes scheduling recurring jobs or automated workflows (e.g., cron-like triggers, scheduled report generation, or recurring automation pipelines). Webhooks are event-driven, not schedule-based, and no docs mention recurring task orchestration.

   ⚑ flag

Calendar

1. product managerI connect Google or Outlook calendar once and control per-meeting-type rules for which meetings get captured automatically

   weight 3 · round to Fellow
   Granola—none0/10

   Evidence shows Granola connects to a calendar to display upcoming meetings and offers templates per call type for note formatting, but there is no mention of per-meeting-type rules controlling which meetings get automatically captured/recorded — no filtering, auto-join rules, or selective-capture settings are documented.

   • [claimed-docs] “Granola connects to your calendar to show your upcoming meetings.”
   • [claimed-docs] “you might want to choose a specific template (or set up your own!) for a specific type of call: for instance, 1:1s, interviews or sales pitc…”
   • [claimed-docs] “you might want to choose a specific template (or set up your own!) for a specific type of call: for instance, 1:1s, interviews or sales pitc…”

   proof ↗⚑ flag

   Fellow~partialclaimed3/10

   Fellow's evidence shows organization-wide recording policies and manual per-meeting record/pause control, but there is no explicit mention of connecting a Google or Outlook calendar or of configuring rules that vary automatic capture by meeting type. Missing for 10: explicit calendar OAuth connection flow, documentation of per-meeting-type (e.g. 1:1 vs external) auto-capture rule configuration, and any UI/API evidence tying calendar events to capture policy.

   • [claimed-docs] “Set organization-wide policies for how meetings are recorded, accessed, and retained”
   • [claimed-docs] “Users choose when to record, pause, or resume capture when part of a conversation needs to stay off the record.”
   • [claimed-docs] “Standardize how meetings are captured, managed, and protected across every platform and format your teams use.”
   • [claimed-docs] “Set configurable retention policies, including zero-day retention, so data is handled according to your organization’s requirements.”

   proof ↗⚑ flag

2. founderI walk into each meeting prepped — the tool surfaces past meetings with the same people and a brief of open threads before the call

   weight 1 · round to Fellow
   Granola~partialclaimed4/10

   Granola stores meeting notes/transcripts and lets users search or chat with past notes (via Chat or MCP's 'search through your meeting notes, find specific topics discussed in past meetings'), and it connects to your calendar to show upcoming meetings, which together could support manual meeting prep. However, there is no evidence of an automatic pre-call brief or a feature that proactively surfaces past meetings with the same attendees before a call starts — the burden is on the user to search/chat manually. Missing for 10: automatic surfacing of past meetings by attendee tied to calendar events, a dedicated pre-meeting briefing/open-threads summary feature, and any first-hand account of this proactive workflow.

   • [claimed-docs] “Granola connects to your calendar to show your upcoming meetings.”
   • [claimed-docs] “Granola Chat lets you ask questions about your meetings and get instant insights from your notes or transcripts.”
   • [claimed-docs] “Connect your AI tools to your Granola meeting notes with Granola MCP. Once connected you can use this to: search through your meeting notes,…”
   • [claimed-docs] “Chat with your meeting notes to recall meeting details, spot patterns across customer calls, and generate follow-up outputs.”

   proof ↗⚑ flag

   Fellow~partialclaimed5/10

   Fellow makes meetings searchable via full transcript search and MCP/API queries like 'what did we decide in last month's standup', which supports surfacing past context with specific people, but there is no documented feature that proactively assembles a pre-meeting brief keyed to attendees or auto-surfaces prior meetings with the same people before a call. missing for 10: a dedicated pre-meeting briefing/prep feature that auto-detects attendees and compiles related past meetings and open action items ahead of the call, plus independent hands-on evidence of this workflow.

   • [claimed-docs] “Every conversation becomes searchable through a full transcript, giving your team instant access to past context whenever they need it.”
   • [claimed-docs] “Ask anything about your meetings - Get answers to questions like "What were my action items this week?" or "What did we decide in last month…”
   • [claimed-docs] “Ask anything about your meetings - Get answers to questions like "What were my action items this week?"”
   • [claimed-docs] “Your favorite AI chatbot like Claude, ChatGPT, or Cursor can now tap directly into your meeting recaps, notes, and action items”
   • [claimed-docs] “pass transcripts to an LLM for tailored recaps and action plans, the API gives you everything you need to make it happen”

   proof ↗⚑ flag

Capture

1. product managerThe notetaker reliably captures Zoom, Google Meet, and Microsoft Teams meetings, with recording where I want it

   weight 2 · round drawn
   Granola✓fullcommunity8/10

   Granola explicitly claims to work with Zoom, Google Meet, Teams and other apps, and records via computer audio rather than a bot joining the call, directly addressing the 'recording where I want it' preference. Community evidence corroborates real-world reliability (battery/performance) and PM usage, though no independent side-by-side accuracy test across all three platforms is present. Missing for 10: independent verification of capture reliability across each specific platform (Zoom/Meet/Teams) and no mention of optional bot-based recording for cases where computer-audio capture isn't desired.

   • [claimed-docs] “Works with Zoom, Google Meet, Teams and every other meeting app.”
   • [claimed-docs] “Uses your computer audio, so doesn't invite a bot”
   • [community] “I've been beta testing it and it seems like it's not bad for battery life at all, works like a charm.”
   • [community] “I work as a PM and have a lot of calls and action items that tend to slip through so I've built a Mac app that turns Granola calls into a ka…”

   proof ↗⚑ flag

   Fellow✓fullclaimed8/10

   Fellow explicitly documents recording capture across Zoom, Google Meet, Microsoft Teams (plus Slack Huddles and in-person), with flexibility to record via visible bot or bot-less desktop audio capture, and user control over when to record/pause/resume. This directly matches the PM's need for reliable, controllable capture across major platforms. Missing for 10: independent/hands-on verification of recording reliability across all three platforms and no third-party reviews confirming real-world capture accuracy.

   • [claimed-docs] “Fellow gives teams the flexibility to record with a visible bot or without one — capturing audio directly from the desktop”
   • [claimed-docs] “Fellow gives teams the flexibility to record with a visible bot or without one — capturing audio directly from the desktop across Zoom, Goog…”
   • [claimed-docs] “Users choose when to record, pause, or resume capture when part of a conversation needs to stay off the record.”
   • [claimed-docs] “Standardize how meetings are captured, managed, and protected across every platform and format your teams use.”

   proof ↗⚑ flag

2. founderCapture and transcribe meetings without a visible bot joining the call — audio is captured from my device so external participants see nothing extra

   weight 3 · round to Granola
   Granola✓fullcommunity9/10

   Granola's core value prop is capturing device/computer audio directly without a bot joining the call, explicitly stated as 'Uses your computer audio, so doesn't invite a bot' and works across Zoom/Meet/Teams/etc, with transparency messaging posted in chat rather than a visible bot presence. Community evidence corroborates it runs quietly in the background on the user's device. Missing for 10: independent hands-on confirmation from a founder-specific use case and no explicit contradicting report of a visible bot ever appearing.

   • [claimed-docs] “Uses your computer audio, so doesn't invite a bot”
   • [claimed-docs] “Works with Zoom, Google Meet, Teams and every other meeting app.”
   • [claimed-docs] “Granola posts a message in your meeting chat when it starts transcribing to let everyone know.”
   • [community] “Granola terrifies me. It’s highly capable at summarisation, and seems to be free. What’s the catch? ... it runs in the background on my Mac,…”

   proof ↗⚑ flag

   Fellow✓fullclaimed8/10

   Fellow explicitly documents botless recording that captures audio directly from the desktop across Zoom, Meet, Teams, Slack Huddles and in-person meetings, giving teams flexibility to record with or without a visible bot, plus transcript generation and search from the captured audio. Missing for 10: independent/hands-on confirmation that external participants truly see nothing extra during botless capture, and detail on how desktop audio capture technically avoids appearing in participant lists.

   • [claimed-docs] “Fellow gives teams the flexibility to record with a visible bot or without one — capturing audio directly from the desktop”
   • [claimed-docs] “Fellow gives teams the flexibility to record with a visible bot or without one — capturing audio directly from the desktop across Zoom, Goog…”
   • [claimed-docs] “Every conversation becomes searchable through a full transcript, giving your team instant access to past context whenever they need it.”
   • [claimed-docs] “Users choose when to record, pause, or resume capture when part of a conversation needs to stay off the record.”

   proof ↗⚑ flag

3. founderRecord and transcribe in-person conversations from a mobile app, and those notes land in the same searchable workspace

   weight 2 · round to Granola
   Granola~partialclaimed6/10

   Granola's docs confirm a mobile app for recording AI-enhanced notes during in-person meetings and iOS calls, and separate docs confirm workspace-wide search/chat over meeting notes, but no evidence explicitly ties mobile-captured notes into the same searchable workspace pipeline. missing for 10: explicit confirmation that mobile-recorded notes/transcripts sync into the same workspace and are searchable via Chat/API alongside desktop notes, and any hands-on user report validating this integration.

   • [claimed-docs] “Granola for mobile lets you take AI-enhanced notes during in-person meetings (and, on iOS, phone calls).”
   • [claimed-docs] “Granola Chat lets you ask questions about your meetings and get instant insights from your notes or transcripts.”
   • [claimed-docs] “Use the magnifying glass beside each note to see where in the transcript or raw notes the note came from”
   • [claimed-docs] “Granola connects to your calendar to show your upcoming meetings.”

   proof ↗⚑ flag

   Fellow—none0/10

   Evidence shows Fellow can capture in-person meetings via desktop audio capture (fellow-docs-20) and makes transcripts searchable (fellow-docs-21), but there is no mention anywhere in the pack of a mobile app for recording conversations — all capture references point to desktop/bot-based capture during scheduled meetings. Missing for 10: any documentation of a Fellow mobile app, mobile recording capability, or evidence that in-person capture happens via phone rather than laptop/desktop.

   • [claimed-docs] “Fellow gives teams the flexibility to record with a visible bot or without one — capturing audio directly from the desktop across Zoom, Goog…”
   • [claimed-docs] “Every conversation becomes searchable through a full transcript, giving your team instant access to past context whenever they need it.”
   • [claimed-docs] “Users choose when to record, pause, or resume capture when part of a conversation needs to stay off the record.”

   proof ↗⚑ flag

Integrations

1. sales leadMeeting notes and summaries sync automatically onto the right contact and deal records in my CRM (HubSpot, Salesforce, Attio)

   weight 3 · round to Granola
   Granola~partialclaimed6/10

   Granola has a documented native HubSpot integration that auto-syncs meeting notes to the correct contacts, companies, and deals via folders (granola-docs-10, granola-docs-39), which directly matches the story for HubSpot. However, there is no dedicated Salesforce or Attio integration documented — those CRMs would only be reachable indirectly via the generic Zapier connector (granola-docs-9), not a first-party record-matching sync. Missing for 10: explicit native Salesforce and Attio CRM sync with contact/deal matching, and independent/hands-on confirmation that the HubSpot sync reliably attaches to the 'right' record.

   • [claimed-docs] “Sync Granola meeting notes to HubSpot. Automatically sync notes to CRM records from your meetings using folders.”
   • [claimed-docs] “Automatically sync notes to CRM records from your meetings using folders.”
   • [claimed-docs] “You can set up Granola to automatically save any notes that are added to Granola folders to the relevant contacts, companies, or deals on Hu…”
   • [claimed-docs] “Connect Granola to 8,000+ apps and automate what happens after a meeting”

   proof ↗⚑ flag

   Fellow—none0/10

   The evidence describes Fellow's Developer API, webhooks, and MCP server for accessing transcripts/notes/action items, but there is no mention of any native integration or automatic sync with HubSpot, Salesforce, or Attio contact/deal records. A sales-focused meeting tool could plausibly ship such CRM sync, so the axis applies, but no evidence confirms it exists.

   • [claimed-docs] “The Developer API opens that data to you: transcripts, structured notes, and more, all available through standard REST calls.”
   • [claimed-docs] “Fellow sends an HTTP POST request to the URL you specify with details about the event.”
   • [claimed-docs] “Webhooks allow you to receive real-time notifications when events occur in Fellow.”
   • [claimed-docs] “pass transcripts to an LLM for tailored recaps and action plans, the API gives you everything you need to make it happen”

   proof ↗⚑ flag

2. product managerNotes flow into the tools where work happens — Slack channels, Notion pages, and thousands of apps via Zapier

   weight 2 · round to Granola
   Granola✓fullclaimed8/10

   Granola explicitly documents Zapier integration to 8,000+ apps, native HubSpot CRM sync, and Chat workflows that draft/send Slack messages and follow-up emails, directly matching the story's 'flows into Slack, Notion, and thousands of apps' vision. Notion isn't named directly, but it would be covered under the 8,000+ app Zapier ecosystem; missing for 10: an explicit first-party Notion integration/doc and independent user confirmation that Slack/Notion syncing works reliably in practice.

   • [claimed-docs] “Connect Granola to 8,000+ apps and automate what happens after a meeting”
   • [claimed-docs] “Sync Granola meeting notes to HubSpot. Automatically sync notes to CRM records from your meetings using folders.”
   • [claimed-docs] “Automatically sync notes to CRM records from your meetings using folders.”
   • [claimed-docs] “draft and send a follow-up email" ... "draft and send a Slack message" ... "find a time and create a Google Calendar event”
   • [claimed-docs] “Ask Chat to: draft and send a follow-up email * draft and send a Slack message * find a time and create a Google Calendar event”
   • [claimed-docs] “You can set up Granola to automatically save any notes that are added to Granola folders to the relevant contacts, companies, or deals on Hu…”

   proof ↗⚑ flag

   Fellow—none0/10

   The evidence pack only documents a generic Developer API, webhooks, and MCP server — there is no mention of native Slack, Notion, or Zapier integrations or an app marketplace listing that would let notes flow automatically into those specific tools.

   ⚑ flag

Memory

1. product managerAsk questions in natural language across my whole meeting history ("what did we decide about pricing?") and get answers with sources

   weight 2 · round to Granola
   Granola✓fullprobed7/10

   Granola Chat explicitly lets users ask natural-language questions across meeting notes/transcripts and get insights, with a magnifying-glass feature to trace an answer back to its source location in the transcript/notes, and the MCP/API integrations further support searching past meetings and surfacing sourced context. missing for 10: independent/hands-on validation of answer accuracy and citation quality, and explicit confirmation that native Chat (not just MCP) searches the *entire* historical corpus rather than per-meeting.

   • [claimed-docs] “Granola Chat lets you ask questions about your meetings and get instant insights from your notes or transcripts.”
   • [claimed-docs] “Chat with your meeting notes to recall meeting details, spot patterns across customer calls, and generate follow-up outputs.”
   • [claimed-docs] “Use the magnifying glass beside each note to see where in the transcript or raw notes the note came from”
   • [claimed-docs] “Connect your AI tools to your Granola meeting notes with Granola MCP. Once connected you can use this to: search through your meeting notes,…”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to the official hosted MCP server https://mcp.granola.ai/mcp returned …”

   proof ↗⚑ flag

   Fellow~partialprobed7/10

   Fellow's MCP server explicitly supports natural-language queries across meeting history (e.g. 'What did we decide in last month's standup?') and searchable full transcripts back the answers, with a runtime probe confirming the hosted MCP endpoint is live and functional. However, the evidence never explicitly describes source citations/links accompanying answers, only that transcripts and notes are the underlying data. Missing for 10: explicit documentation that MCP answers include citations/sources to specific meetings, and independent hands-on confirmation of answer quality/accuracy.

   • [claimed-docs] “Ask anything about your meetings - Get answers to questions like "What were my action items this week?"”
   • [claimed-docs] “Fellow's Model Context Protocol (MCP) Server allows you to ask your AI Assistant about your meetings, without needing to write any code.”
   • [claimed-docs] “Ask anything about your meetings - Get answers to questions like "What were my action items this week?" or "What did we decide in last month…”
   • [claimed-docs] “Every conversation becomes searchable through a full transcript, giving your team instant access to past context whenever they need it.”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to Fellow's hosted MCP server https://fellow.app/mcp returned HTTP 401…”
   • [probe] “official MCP server documented at https://developers.fellow.ai/reference/mcp-server”

   proof ↗⚑ flag

2. product managerSearch across every past meeting — transcripts, summaries, and notes — and jump to the exact moment something was said

   weight 3 · round to Granola
   Granola✓fullprobed8/10

   Granola Chat lets users ask questions and get insights across notes/transcripts (docs-19, docs-4), the MCP server explicitly supports searching through meeting notes and finding topics across past meetings plus retrieving full transcripts (docs-35, granola-probe-rt-1), and the in-app magnifying-glass feature jumps to the exact transcript/note location a note came from (docs-27). Missing for 10: independent/hands-on evidence confirming true cross-meeting semantic search quality (vs. single-note context) and no explicit documentation of a dedicated 'search all meetings' UI beyond Chat/MCP.

   • [claimed-docs] “Granola Chat lets you ask questions about your meetings and get instant insights from your notes or transcripts.”
   • [claimed-docs] “Connect your AI tools to your Granola meeting notes with Granola MCP. Once connected you can use this to: search through your meeting notes,…”
   • [claimed-docs] “Use the magnifying glass beside each note to see where in the transcript or raw notes the note came from”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to the official hosted MCP server https://mcp.granola.ai/mcp returned …”
   • [claimed-docs] “Chat with your meeting notes to recall meeting details, spot patterns across customer calls, and generate follow-up outputs.”

   proof ↗⚑ flag

   Fellow✓fullprobed7/10

   Fellow explicitly advertises full-transcript search ('Every conversation becomes searchable through a full transcript') and its MCP server exposes tools like search_meetings and get_meeting_transcript for asking natural-language questions like 'what did we decide in last month's standup', confirmed live via a runtime probe. Missing for 10: explicit UI evidence of clicking a search result to jump to the exact timestamp/moment in a recording, and independent (non-vendor) confirmation of search accuracy across large history.

   • [claimed-docs] “Every conversation becomes searchable through a full transcript, giving your team instant access to past context whenever they need it.”
   • [claimed-docs] “Ask anything about your meetings - Get answers to questions like "What were my action items this week?"”
   • [claimed-docs] “Ask anything about your meetings - Get answers to questions like "What were my action items this week?" or "What did we decide in last month…”
   • [claimed-docs] “Your favorite AI chatbot like Claude, ChatGPT, or Cursor can now tap directly into your meeting recaps, notes, and action items”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to Fellow's hosted MCP server https://fellow.app/mcp returned HTTP 401…”
   • [probe] “official MCP server documented at https://developers.fellow.ai/reference/mcp-server”

   proof ↗⚑ flag

Summaries

1. product managerAction items with owners are automatically extracted from the conversation and collected somewhere I can track them

   weight 3 · round to Fellow
   Granola—none0/10

   Granola's docs describe general AI summaries, chat Q&A, and workflow actions (draft follow-up emails, Slack messages, calendar events) but never mention structured action-item extraction with assigned owners or a dedicated tracking view/list. Community evidence reinforces this gap: a PM explicitly built a separate third-party app to turn Granola meeting notes into a kanban to-do list because action items 'tend to slip through' natively.

   • [claimed-docs] “Granola Chat can help you complete the work that comes after a meeting... draft and send a follow-up email”
   • [claimed-docs] “draft and send a follow-up email" ... "draft and send a Slack message" ... "find a time and create a Google Calendar event”
   • [claimed-docs] “Ask Chat to: draft and send a follow-up email * draft and send a Slack message * find a time and create a Google Calendar event”
   • [community] “I work as a PM and have a lot of calls and action items that tend to slip through so I've built a Mac app that turns Granola calls into a ka…”

   proof ↗⚑ flag

   Fellow✓fullclaimed8/10

   Fellow automatically generates AI action items with assignees from meeting conversations (fellow-docs-23), and these are collected/trackable via the app and API's get_action_items endpoint with filters like assigned_to_me/assigned_to_others (fellow-docs-5, fellow-docs-19), plus surfaced via MCP queries like 'what were my action items' (fellow-docs-32). Missing for 10: no independent/hands-on evidence confirming extraction accuracy or owner-assignment reliability beyond vendor docs.

   • [claimed-docs] “You automatically receive a clear meeting summary, suggested AI action items, decisions, and topic-based minutes for every meeting.”
   • [claimed-docs] “List action items with optional filters and pagination.”
   • [claimed-docs] “assigned_to_me: Only action items assigned to the authenticated user\n- assigned_to_others: Only action items the user can edit, assigned to…”
   • [claimed-docs] “Ask anything about your meetings - Get answers to questions like "What were my action items this week?" or "What did we decide in last month…”
   • [claimed-docs] “pass transcripts to an LLM for tailored recaps and action plans, the API gives you everything you need to make it happen”

   proof ↗⚑ flag

2. sales leadShape the notes with custom templates or saved prompts per meeting type (discovery call, 1:1, standup) instead of one generic format

   weight 2 · round to Granola
   Granola✓fullclaimed8/10

   Granola explicitly supports custom/default templates per call type (1:1s, interviews, sales pitches) and 'Recipes' as reusable prompt templates to avoid retyping the same questions, directly matching the story's need for per-meeting-type customization. Missing for 10: independent/hands-on user corroboration of template customization quality and confirmation that saved 'recipes' can be tied to specific meeting types automatically.

   • [claimed-docs] “you might want to choose a specific template (or set up your own!) for a specific type of call: for instance, 1:1s, interviews or sales pitc…”
   • [claimed-docs] “you might want to choose a specific template (or set up your own!) for a specific type of call: for instance, 1:1s, interviews or sales pitc…”
   • [claimed-docs] “While Granola's default summarisation adapts to a range of formats, you might want to choose a specific template (or set up your own!) for a…”
   • [claimed-docs] “Recipes are reusable prompt templates that help you quickly get insights from your meeting notes without typing the same questions repeatedl…”

   proof ↗⚑ flag

   Fellow—none0/10

   The evidence pack covers Fellow's API, MCP server, webhooks, recording controls, and retention policies, but contains no mention of customizable note templates or saved prompts tied to meeting type (discovery call, 1:1, standup). This is a fair axis for a meeting-notes product, but no evidence supports it.

   ⚑ flag

3. sales leadThe tool drafts my follow-up email from the meeting so I can review and send it in a couple of clicks

   weight 1 · round to Granola
   Granola✓fullclaimed8/10

   Granola Chat explicitly supports drafting and sending a follow-up email directly from meeting notes as a documented workflow, with sales-relevant templates (e.g., sales pitch note templates) and CRM sync options for follow-through. Missing for 10: no independent/hands-on review confirming the 'couple of clicks' review-and-send UX in practice, and no detail on how much editing/review is typically needed before sending.

   • [claimed-docs] “Granola Chat can help you complete the work that comes after a meeting... draft and send a follow-up email”
   • [claimed-docs] “draft and send a follow-up email" ... "draft and send a Slack message" ... "find a time and create a Google Calendar event”
   • [claimed-docs] “Ask Chat to: draft and send a follow-up email * draft and send a Slack message * find a time and create a Google Calendar event”
   • [claimed-docs] “Granola Chat can help you complete the work that comes after a meeting, without switching between your notes and other apps.”
   • [claimed-docs] “you might want to choose a specific template (or set up your own!) for a specific type of call: for instance, 1:1s, interviews or sales pitc…”

   proof ↗⚑ flag

   Fellow—none0/10

   Fellow's evidence covers meeting summaries, action items, transcripts, and an API/MCP server for querying meeting data and even passing transcripts to an LLM for 'tailored recaps and action plans,' but there is no evidence of a native feature that drafts a follow-up email for one-click review and send — that would require custom API integration, not a documented built-in capability.

   • [claimed-docs] “You automatically receive a clear meeting summary, suggested AI action items, decisions, and topic-based minutes for every meeting.”
   • [claimed-docs] “pass transcripts to an LLM for tailored recaps and action plans, the API gives you everything you need to make it happen”
   • [claimed-docs] “Ask anything about your meetings - Get answers to questions like "What were my action items this week?"”
   • [claimed-docs] “Ask anything about your meetings - Get answers to questions like "What were my action items this week?" or "What did we decide in last month…”

   proof ↗⚑ flag

4. product managerI get a structured summary right after each meeting — key points, decisions, and next steps — good enough to share without editing

   weight 3 · round to Fellow
   Granola✓fullclaimed7/10

   Granola's core AI-enhanced notes produce structured summaries with templates for different meeting types (1:1s, interviews, sales pitches), speaker attribution, and searchable source-linked notes, which are shareable via link or CSV export without further editing. missing for 10: independent hands-on evaluation of summary quality/accuracy (e.g., a reviewer confirming it required no edits), and no direct evidence addressing 'decisions' and 'next steps' as distinct structured fields versus a general summary format.

   • [claimed-docs] “you might want to choose a specific template (or set up your own!) for a specific type of call: for instance, 1:1s, interviews or sales pitc…”
   • [claimed-docs] “you might want to choose a specific template (or set up your own!) for a specific type of call: for instance, 1:1s, interviews or sales pitc…”
   • [claimed-docs] “While Granola's default summarisation adapts to a range of formats, you might want to choose a specific template (or set up your own!) for a…”
   • [claimed-docs] “Speaker tags show each participant's name above their parts of the conversation, giving you more detailed attribution than the default Me an…”
   • [claimed-docs] “Use the magnifying glass beside each note to see where in the transcript or raw notes the note came from”
   • [claimed-docs] “Copy link will copy a unique URL link to the meeting notes, which you can send to others.”
   • [claimed-docs] “The CSV includes the title, note summary, transcript, and other basic details for each note.”

   proof ↗⚑ flag

   Fellow✓fullclaimed8/10

   Fellow's core product docs state you automatically receive a clear meeting summary, suggested AI action items, decisions, and topic-based minutes for every meeting, directly matching the story of a structured post-meeting summary. Missing for 10: independent/hands-on user reviews confirming the summary quality is share-ready without edits, and no detail on editing workflow before sharing.

   • [claimed-docs] “You automatically receive a clear meeting summary, suggested AI action items, decisions, and topic-based minutes for every meeting.”
   • [claimed-docs] “Every conversation becomes searchable through a full transcript, giving your team instant access to past context whenever they need it.”
   • [claimed-docs] “Fellow gives teams the flexibility to record with a visible bot or without one — capturing audio directly from the desktop”

   proof ↗⚑ flag

1. ai-native userDo everything through the API that I can do in the UI

   weight 2 · round to Fellow
   Granola~partialprobed4/10

   The Granola API and MCP server give programmatic read access to meeting notes, transcripts, and summaries, and support building custom agents (granola-docs-1/16/34/35, granola-probe-rt-1 confirms live OAuth MCP endpoint), but there's no documented API endpoint for the many interactive UI capabilities like Chat workflows (draft/send email, Slack, calendar events), Recipes/templates, folder-based CRM sync, or transcript auto-deletion settings — these remain UI-only or handled via separate third-party integrations (Zapier/HubSpot). No OpenAPI spec was found (granola-probe-2 shows 404s), suggesting a narrow, undocumented surface rather than full UI parity. Missing for 10: API endpoints for chat-driven actions, template/recipe management via API, and any documented parity for sharing/export settings.

   • [claimed-docs] “The Granola API provides programmatic access to your workspace's meeting notes and related data.”
   • [claimed-docs] “The Granola API lets you build custom integrations with programmatic access to meeting notes, transcripts, and AI summaries.”
   • [claimed-docs] “Custom AI agents - Build agents that analyze sentiment across your calls, surface recurring themes in product feedback, or combine meeting c…”
   • [claimed-docs] “Connect your AI tools to your Granola meeting notes with Granola MCP. Once connected you can use this to: search through your meeting notes,…”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless JSON-RPC initialize POST to the official hosted MCP server https://mcp.granola.ai/mcp returned …”
   • [probe] “PROBE openapi: all candidate paths 404 (https://docs.granola.ai/openapi.json, https://docs.granola.ai/swagger.json, https://docs.granola.ai/…”
   • [claimed-docs] “Ask Chat to: draft and send a follow-up email * draft and send a Slack message * find a time and create a Google Calendar event”
   • [claimed-docs] “Recipes are reusable prompt templates that help you quickly get insights from your meeting notes without typing the same questions repeatedl…”

   proof ↗⚑ flag

   Fellow~partialprobed5/10

   Fellow's Developer API exposes meaningful UI-parallel functionality (list recordings, list action items, append note agenda, webhooks, MCP conversational access) but the evidence only covers a subset of read/limited-write operations, not full parity with UI actions like recording controls, retention policy configuration, keyword tracking, or admin dashboard settings. missing for 10: API coverage for recording start/pause/resume, retention/compliance policy management via API, keyword tracking configuration, and broader write/update operations beyond appending agenda text.

   • [claimed-docs] “The Developer API opens that data to you: transcripts, structured notes, and more, all available through standard REST calls.”
   • [claimed-docs] “List recordings with optional filters and pagination.”
   • [claimed-docs] “List action items with optional filters and pagination.”
   • [claimed-docs] “Add Fellow Markdown to the end of a note's agenda, leaving the rest of the content in place.”
   • [claimed-docs] “Webhooks allow you to receive real-time notifications when events occur in Fellow.”
   • [claimed-docs] “pass transcripts to an LLM for tailored recaps and action plans, the API gives you everything you need to make it happen”
   • [probe] “PROBE openapi: all candidate paths 404 (https://developers.fellow.ai/openapi.json, https://developers.fellow.ai/swagger.json, https://develo…”

   proof ↗⚑ flag

2. ai-native userExport all of my data in open formats and leave

   weight 3 · round to Fellow
   Granola~partialcommunity5/10

   Granola documents a CSV export of meeting notes (titles, summaries, transcripts) via Settings, plus API/MCP access to notes data, giving a real open-format export path, but this covers only notes/transcripts, not a full account/data export (templates, folders, integration settings, chat history) that would let a user fully leave with everything. Community commentary also frames export as a manual workaround ('copy-paste note summaries out to Obsidian') rather than a first-class 'export everything' feature. Missing for 10: comprehensive full-account export beyond notes/transcripts, and any independent confirmation that exported CSV/API data is truly complete and portable.

   • [claimed-docs] “The CSV includes the title, note summary, transcript, and other basic details for each note.”
   • [claimed-docs] “Export your Granola meeting notes to CSV. Get note titles, summaries, transcripts, and details.”
   • [claimed-docs] “You're able to export historical data from Granola by going to Settings, clicking Profile in the sidebar (under Preferences), and clicking G…”
   • [claimed-docs] “This doc walks through how you can export your meeting notes from Granola, including transcripts.”
   • [claimed-docs] “The Granola API lets you build custom integrations with programmatic access to meeting notes, transcripts, and AI summaries.”
   • [community] “It's free for a limited number of notes. I guess you could copypaste the note summaries out to Obsidian or something, but at that point you'…”

   proof ↗⚑ flag

   Fellow~partialclaimed6/10

   Fellow's Developer API exposes transcripts, notes, action items, and recordings via standard REST calls, and the Super Admin API can 'retrieve, export, and delete data across the entire workspace' for Enterprise admins, giving a path to full data export. However, there is no explicit documentation of a bulk 'export everything' feature, no stated output format guarantees (e.g., JSON/CSV schema for full account export), and no independent verification of completeness. Missing for 10: a documented one-click/bulk full-account export tool, explicit open-format (JSON/CSV) export spec, and independent confirmation that all data types are covered.

   • [claimed-docs] “The Developer API opens that data to you: transcripts, structured notes, and more, all available through standard REST calls.”
   • [claimed-docs] “List recordings with optional filters and pagination.”
   • [claimed-docs] “List action items with optional filters and pagination.”
   • [claimed-docs] “Fellow’s Super Admin API lets designated Enterprise‑plan administrators retrieve, export, and delete data across the entire workspace”
   • [claimed-docs] “pass transcripts to an LLM for tailored recaps and action plans, the API gives you everything you need to make it happen”

   proof ↗⚑ flag

Pricing

1. founderStart free and see exactly what each paid tier costs and adds — no "talk to sales" wall for basic use

   weight 2 · round drawn
   Granola—none0/10

   No evidence pack content addresses Granola's pricing tiers, free plan limits, or paid plan costs/features; only community comments vaguely mention 'free for a limited number of notes' without detail on tier costs or sales-contact requirements. missing for 10: pricing page details, tier breakdown, cost transparency documentation, confirmation of no 'talk to sales' wall.

   • [community] “It's free for a limited number of notes. I guess you could copypaste the note summaries out to Obsidian or something, but at that point you'…”

   proof ↗⚑ flag

   Fellow—none0/10

   The evidence pack only mentions two feature names ('Keyword tracking', 'Transcript redaction') from the pricing page but contains no details on free tier availability, tier pricing, or whether higher tiers require contacting sales. No evidence supports or contradicts a transparent self-serve pricing structure.

   • [claimed-docs] “Keyword tracking”
   • [claimed-docs] “Transcript redaction”

   proof ↗⚑ flag

Privacy

1. it adminThe product ships real consent features — participant notifications, in-meeting disclosure, or admin-enforced transparency — not just a policy PDF

   weight 3 · round to Granola
   Granola~partialclaimed6/10

   Granola documents concrete in-meeting disclosure (a chat message announcing transcription start) and consent-adjacent controls like opting out of model training and auto-deleting transcripts, going beyond a mere policy PDF. However, there is no evidence of admin-enforced transparency settings (e.g., workspace-wide consent policy enforcement, participant opt-out controls, or audit of consent compliance) that would let an IT admin mandate or verify these behaviors across the org. Missing for 10: admin-level enforcement/configuration of consent notifications, participant-side consent management, and any audit trail for consent compliance.

   • [claimed-docs] “Granola posts a message in your meeting chat when it starts transcribing to let everyone know.”
   • [claimed-docs] “You can opt out of model training in Settings.”
   • [claimed-docs] “Granola allows you to automatically delete transcripts after a set period of time.”
   • [claimed-docs] “Granola allows you to automatically delete transcripts after a set period of time. This applies only to transcripts — your notes are not aff…”
   • [claimed-docs] “Your notes are never shared with any participants unless you share the link, or add the note to a folder in your team space.”

   proof ↗⚑ flag

   Fellow—none0/10

   The evidence shows admin-side policy controls (retention settings, org-wide recording/access policies) and user-side recording controls, but nothing describing participant-facing consent mechanisms such as in-meeting recording disclosure banners, automated participant notifications when a bot joins, or admin-enforced transparency toward attendees. Notably, Fellow explicitly supports recording 'without a visible bot' by capturing desktop audio directly, which underscores the absence of a documented notification/disclosure mechanism rather than confirming one.

   • [claimed-docs] “Fellow gives teams the flexibility to record with a visible bot or without one — capturing audio directly from the desktop”
   • [claimed-docs] “Fellow gives teams the flexibility to record with a visible bot or without one — capturing audio directly from the desktop across Zoom, Goog…”
   • [claimed-docs] “Users choose when to record, pause, or resume capture when part of a conversation needs to stay off the record.”
   • [claimed-docs] “Set organization-wide policies for how meetings are recorded, accessed, and retained”
   • [claimed-docs] “Set configurable retention policies, including zero-day retention, so data is handled according to your organization’s requirements.”
   • [claimed-docs] “Standardize how meetings are captured, managed, and protected across every platform and format your teams use.”

   proof ↗⚑ flag

2. it adminThe vendor documents whether meeting data trains AI models and gives my org an enforceable opt-out, alongside SOC 2 / HIPAA posture

   weight 2 · round to Granola
   Granola~partialclaimed4/10

   Granola's docs confirm an opt-out setting for model training (granola-docs-26) and note transparency/consent features like auto-deletion of transcripts and start-of-recording notifications (granola-docs-12, granola-docs-25), but there is no evidence in the pack of SOC 2 or HIPAA compliance documentation, a formal enterprise DPA, or admin-enforced (org-wide) opt-out controls as opposed to per-user settings. Missing for 10: SOC 2/HIPAA certification pages, org-level enforceable opt-out (vs individual toggle), and any compliance/trust-center documentation.

   • [claimed-docs] “You can opt out of model training in Settings.”
   • [claimed-docs] “Granola allows you to automatically delete transcripts after a set period of time.”
   • [claimed-docs] “Granola posts a message in your meeting chat when it starts transcribing to let everyone know.”

   proof ↗⚑ flag

   Fellow—none0/10

   Evidence shows retention policies and org-wide recording/access controls, but there is no documentation addressing whether meeting data is used to train AI models, no explicit opt-out mechanism for AI training, and no mention of SOC 2 or HIPAA compliance posture anywhere in the pack.

   • [claimed-docs] “Set organization-wide policies for how meetings are recorded, accessed, and retained”
   • [claimed-docs] “Set configurable retention policies, including zero-day retention, so data is handled according to your organization’s requirements.”
   • [claimed-docs] “Standardize how meetings are captured, managed, and protected across every platform and format your teams use.”

   proof ↗⚑ flag

3. it adminSet retention policies — auto-delete transcripts and recordings on a schedule — and permanently erase data on demand

   weight 2 · round to Fellow
   Granola~partialclaimed5/10

   Docs confirm Granola supports automatic deletion of transcripts after a set period (a scheduled retention policy), but this is framed as a general privacy setting, not clearly an IT-admin-controlled workspace policy, and notes/recordings are explicitly excluded ('your notes are not affected'). There is no evidence of an on-demand 'permanently erase all data' admin action or coverage of audio recordings specifically. Missing for 10: admin-level retention policy management console, recordings deletion (not just transcripts), and an on-demand permanent erase/purge capability.

   • [claimed-docs] “Granola allows you to automatically delete transcripts after a set period of time.”
   • [claimed-docs] “Granola allows you to automatically delete transcripts after a set period of time. This applies only to transcripts — your notes are not aff…”

   proof ↗⚑ flag

   Fellow~partialclaimed6/10

   Fellow documents configurable retention policies including zero-day retention, and the Super Admin API allows retrieving, exporting, and deleting data across the workspace, which supports on-demand erasure. However, there's no explicit documentation of a scheduled auto-delete mechanism for transcripts/recordings specifically or a UI/API for setting retention schedules with confirmation of enforcement. Missing for 10: detailed retention policy configuration UI/API docs, evidence of scheduled automated deletion execution, and independent confirmation that erasure requests are honored end-to-end.

   • [claimed-docs] “Set configurable retention policies, including zero-day retention, so data is handled according to your organization’s requirements.”
   • [claimed-docs] “Fellow’s Super Admin API lets designated Enterprise‑plan administrators retrieve, export, and delete data across the entire workspace”
   • [claimed-docs] “Set organization-wide policies for how meetings are recorded, accessed, and retained”
   • [claimed-docs] “Standardize how meetings are captured, managed, and protected across every platform and format your teams use.”

   proof ↗⚑ flag

1. ai-native userChoose where my data is stored (region/residency)

   weight 2 · round drawn
   Granola—none0/10

   No evidence anywhere in the docs or community mentions data residency, regional storage options, or geographic control over where meeting notes/transcripts are stored; only auto-deletion and opt-out-of-training controls are documented, which is a different privacy axis.

   • [claimed-docs] “Granola allows you to automatically delete transcripts after a set period of time.”
   • [claimed-docs] “You can opt out of model training in Settings.”
   • [claimed-docs] “Granola allows you to automatically delete transcripts after a set period of time. This applies only to transcripts — your notes are not aff…”

   proof ↗⚑ flag

   Fellow—none0/10

   No evidence pack item mentions data residency, regional data storage options, or geographic data center selection; retention policies and admin controls are covered but region choice is not addressed.

   ⚑ flag

2. ai-native userPrevent my data from being used to train AI models

   weight 3 · round to Granola
   Granola✓fullclaimed8/10

   Granola docs explicitly state users can opt out of model training via Settings, directly delivering the story's core requirement. Missing for 10: independent/hands-on confirmation that opt-out actually stops training, and details on scope (e.g., whether opted-out data is retroactively excluded).

   • [claimed-docs] “You can opt out of model training in Settings.”

   proof ↗⚑ flag

   Fellow—none0/10

   The evidence pack covers retention policies, redaction, recording controls, and API/MCP access, but contains no statement about whether Fellow uses customer data to train AI models or any opt-out mechanism for AI training.

   ⚑ flag

3. ai-native userControl data retention and deletion

   weight 2 · round to Fellow
   Granola~partialclaimed5/10

   Docs confirm concrete retention controls: automatic transcript deletion after a set period (transcripts only, notes unaffected) and an opt-out of model training in Settings, giving users some control over data handling. However, there's no evidence of full note/data deletion, account-level data purge, or granular retention policies beyond transcripts. Missing for 10: full note/data deletion controls, account-wide data export-then-delete workflow, and documented retention periods/config options beyond transcripts.

   • [claimed-docs] “Granola allows you to automatically delete transcripts after a set period of time.”
   • [claimed-docs] “Granola allows you to automatically delete transcripts after a set period of time. This applies only to transcripts — your notes are not aff…”
   • [claimed-docs] “You can opt out of model training in Settings.”

   proof ↗⚑ flag

   Fellow✓fullclaimed8/10

   Fellow documents configurable retention policies including zero-day retention, org-wide policies on how data is recorded/accessed/retained, and Super Admin API to retrieve, export, and delete data across the entire workspace, directly addressing retention/deletion control. missing for 10: independent/hands-on verification of deletion actually purging data end-to-end and clearer per-user (non-admin) self-service deletion controls.

   • [claimed-docs] “Set configurable retention policies, including zero-day retention, so data is handled according to your organization’s requirements.”
   • [claimed-docs] “Set organization-wide policies for how meetings are recorded, accessed, and retained”
   • [claimed-docs] “Fellow’s Super Admin API lets designated Enterprise‑plan administrators retrieve, export, and delete data across the entire workspace”
   • [claimed-docs] “Standardize how meetings are captured, managed, and protected across every platform and format your teams use.”

   proof ↗⚑ flag

4. ai-native userOpt out of telemetry and usage tracking

   weight 2 · round to Granola
   Granola~partialclaimed5/10

   Granola offers an explicit opt-out of model training in Settings, which is a direct telemetry/data-usage control, but there is no evidence of a broader opt-out for general product usage/analytics telemetry beyond this single toggle. missing for 10: documentation of a general analytics/telemetry opt-out (not just model-training), independent confirmation the opt-out is honored, and any UI/settings detail beyond the one-line mention.

   • [claimed-docs] “You can opt out of model training in Settings.”

   proof ↗⚑ flag

   Fellow—none0/10

   No evidence pack item discusses telemetry opt-out or usage tracking controls for AI-native users; retention/recording policies are mentioned but not telemetry/usage-tracking opt-out specifically.

   ⚑ flag

Sharing

1. sales leadCut a soundbite or highlight clip from a call and share that moment instead of the whole recording

   weight 1 · round drawn
   Granola—none0/10

   Granola's sharing features (copy note link, CSV export, folder sync to CRM) all share the full note/transcript, not a trimmed audio/video soundbite or highlight clip; no evidence anywhere in the docs pack of a clip-cutting or highlight-extraction feature. Missing for 10: any mention of clip/soundbite creation, timestamp-based excerpting, or audio/video snippet sharing.

   • [claimed-docs] “Your notes are never shared with any participants unless you share the link, or add the note to a folder in your team space.”
   • [claimed-docs] “Copy link will copy a unique URL link to the meeting notes, which you can send to others.”
   • [claimed-docs] “The CSV includes the title, note summary, transcript, and other basic details for each note.”
   • [claimed-docs] “This doc walks through how you can export your meeting notes from Granola, including transcripts.”

   proof ↗⚑ flag

   Fellow—none0/10

   Evidence covers transcripts, recordings listing, action items, webhooks, and MCP access, but nothing describes clipping/highlighting a soundbite from a call or sharing a short excerpt instead of the full recording.

   ⚑ flag

2. it adminSharing is controlled — private-by-default notes, granular link/folder/workspace permissions, and admin visibility into what's shared

   weight 2 · round drawn
   Granola~partialclaimed5/10

   Docs confirm private-by-default notes (not shared unless you share a link or add to a team folder) and simple link/folder sharing controls, plus workspace-level API keys for admins, but there is no evidence of granular per-link or per-folder permission settings, workspace-wide sharing policy controls, or an admin dashboard/audit view into what has been shared. Missing for 10: granular link/folder/workspace permission settings, admin visibility/audit log of shared notes, org-wide sharing policy controls.

   • [claimed-docs] “Your notes are never shared with any participants unless you share the link, or add the note to a folder in your team space.”
   • [claimed-docs] “Copy link will copy a unique URL link to the meeting notes, which you can send to others.”
   • [claimed-docs] “You can set up Granola to automatically save any notes that are added to Granola folders to the relevant contacts, companies, or deals on Hu…”
   • [claimed-docs] “Workspace admins on Business and Enterprise plans can also create workspace API keys. These belong to the workspace rather than an individua…”

   proof ↗⚑ flag

   Fellow~partialclaimed5/10

   Fellow offers org-wide retention/recording policies, workspace admin control over MCP tool access, and Super Admin API for workspace-wide visibility, which addresses some admin oversight needs. However, there is no explicit evidence of private-by-default note settings, granular link/folder-level sharing permissions, or a dedicated sharing-audit dashboard for admins. missing for 10: private-by-default note defaults, granular link/folder/workspace permission controls, admin visibility/audit log specifically for shared content.

   • [claimed-docs] “As a workspace admin, you decide which MCP tools your members’ AI assistants can use.”
   • [claimed-docs] “Set organization-wide policies for how meetings are recorded, accessed, and retained”
   • [claimed-docs] “Fellow’s Super Admin API lets designated Enterprise‑plan administrators retrieve, export, and delete data across the entire workspace”
   • [claimed-docs] “Build admin dashboards that display data from a specific user's perspective”
   • [claimed-docs] “Set configurable retention policies, including zero-day retention, so data is handled according to your organization’s requirements.”
   • [claimed-docs] “Standardize how meetings are captured, managed, and protected across every platform and format your teams use.”

   proof ↗⚑ flag

Transcription

1. founderRun meetings in languages other than English — transcription and summaries support many languages and handle language switching

   weight 2 · round to Granola
   Granola~partialclaimed5/10

   Docs explicitly state Granola supports multiple languages for both transcription and generated summaries (granola-docs-11), directly confirming the core claim. However, there is no detail on which languages are supported, no mention of handling mid-meeting language switching, and no independent/hands-on corroboration of multilingual accuracy. missing for 10: list of supported languages, evidence of accurate language-switching handling, independent verification of non-English transcription quality.

   • [claimed-docs] “Granola supports multiple languages for transcription and generated meeting summaries.”

   proof ↗⚑ flag

   Fellow—none0/10

   No evidence in the pack mentions multi-language support, non-English transcription, or handling language switching within meetings; all citations focus on API access, MCP integration, and English-language feature descriptions.

   ⚑ flag

2. product managerThe transcript attributes words to the right named speakers (diarization plus real name matching), not just "Speaker 1/2"

   weight 3 · round to Granola
   Granola~partialclaimed6/10

   Granola's docs confirm a 'Speaker tags' feature that labels transcript segments with participant names rather than generic 'Speaker 1/2' or 'Me/Them' labels, indicating some diarization+name-matching capability. However, there's no detail on how names are resolved (calendar matching, manual tagging, or true voice diarization) and no independent/community evidence validating accuracy of this attribution in multi-speaker calls. Missing for 10: technical explanation of the diarization method, accuracy benchmarks, and hands-on/independent verification of correct name attribution across speakers.

   • [claimed-docs] “Speaker tags show each participant's name above their parts of the conversation, giving you more detailed attribution than the default Me an…”

   proof ↗⚑ flag

   Fellow—none0/10

   The evidence pack covers API access, webhooks, MCP server, and recording/retention features, but nowhere mentions speaker diarization or matching transcript segments to real named speakers versus generic labels like 'Speaker 1/2'.

   ⚑ flag

3. product managerThe vendor documents how transcription actually works and its quality (models, accuracy claims, known limits) rather than just saying "AI-powered"

   weight 3 · round drawn
   Granola—none0/10

   The evidence pack documents feature-level transcription capabilities (multi-language support, speaker tags, auto-deletion, export) but contains no documentation of the underlying transcription model, accuracy benchmarks, or explicit known limitations (e.g., accent/noise handling, word-error-rate). Community comments raise trust/skepticism but do not confirm or deny transcription documentation quality.

   • [claimed-docs] “Granola supports multiple languages for transcription and generated meeting summaries.”
   • [claimed-docs] “Speaker tags show each participant's name above their parts of the conversation, giving you more detailed attribution than the default Me an…”
   • [claimed-docs] “Works with Zoom, Google Meet, Teams and every other meeting app.”

   proof ↗⚑ flag

   Fellow—none0/10

   The evidence pack covers API access, MCP server, webhooks, and recording/retention features, but contains no documentation of the transcription model used, accuracy benchmarks, or known limitations (e.g., accent/language handling, error rates). This axis clearly applies to a transcription product, but no such technical/accuracy disclosure is evidenced.

   ⚑ flag