See what an agent can do with Fin before you ever sign up. Pick a story: recorded sessions replay real probe-harness transcripts; commands tagged live-capable can re-run against the real endpoint from our edge, right now (▶ run live — the exact same request, live and recorded lines always labeled); the live MCP handshake runs real requests from our edge, right now — including, where the server allows it, one real read-only tool call (bring your own key for auth-gated servers); sandboxed self-drive sessions are designed and gated (docs/TRY-IT.md).

$curl -si https://api.intercom.io/merecorded session — replayed, not live

▋

recorded 2026-09-10 · exit 0 · captured verbatim by our probe harness, secrets redacted · pure-HTTP probe — ▶ run live re-runs it from our edge

Verified integrations

Connections to other tracked products — hover a chip for the verbatim evidence quote behind it.

AsanaProject Management

ChatGPTAI Assistants

ClaudeAI Assistants

HubSpotCRM

JiraProject Management

LinearProject Management

NotionProject Management

Salesforce Sales CloudCRM

ShopifyE-commerce Platforms

StripeOnline Payments

ZapierWorkflow Automation

By theme — the product's score on each story themeBy theme

Agent actions — stories about agent actions in this arenaAgent actionsevidence →

Stories about agent actions in this arena

53.6/100

Agenticness — how well agents can access and operate the productAgenticnessevidence →

How well agents can access and operate the product

41.1/100

Automation depth — how much of the product can run unattendedAutomation depthevidence →

How much of the product can run unattended

18.0/100

Channels languages — stories about channels languages in this arenaChannels languagesevidence →

Stories about channels languages in this arena

70.0/100

Escalation handoff — stories about escalation handoff in this arenaEscalation handoffevidence →

Stories about escalation handoff in this arena

36.0/100

Guardrails safety — stories about guardrails safety in this arenaGuardrails safetyevidence →

Stories about guardrails safety in this arena

25.7/100

Insights analytics — stories about insights analytics in this arenaInsights analyticsevidence →

Stories about insights analytics in this arena

27.0/100

Integrations platform — stories about integrations platform in this arenaIntegrations platformevidence →

Stories about integrations platform in this arena

80.0/100

Knowledge grounding — stories about knowledge grounding in this arenaKnowledge groundingevidence →

Stories about knowledge grounding in this arena

46.0/100

Openness — open source, data portability, and self-hosting storiesOpennessevidence →

Open source, data portability, and self-hosting stories

16.8/100

Pricing economics — stories about pricing economics in this arenaPricing economicsevidence →

Stories about pricing economics in this arena

24.0/100

Privacy posture — data-handling and privacy storiesPrivacy postureevidence →

Data-handling and privacy stories

0.0/100

Resolution quality — stories about resolution quality in this arenaResolution qualityevidence →

Stories about resolution quality in this arena

57.3/100

Testing qa — stories about testing qa in this arenaTesting qaevidence →

Stories about testing qa in this arena

70.0/100

Story verdicts — every judged story with its evidenceStory verdicts

What’s free: 0 free · 1 paid · 0 enterprise · 36 not stated in evidence

Sorted by importance (agentic first) (high → low) · 53/53 stories · click a row’s chevron for the rationale and evidence


Connect an agent via an official MCP server G Agent access	ai-native user	Agenticness — how well agents can access and operate the productAgenticness	3	full	8/10	T probed
Delegate tasks to a built-in AI assistant inside the product G Agentic features	ai-native user	Agenticness — how well agents can access and operate the productAgenticness	3	full	8/10	X community
Drive the product through a documented public API G Agent access	ai-native user	Agenticness — how well agents can access and operate the productAgenticness	3	partial	7/10	T probed
Plug MCP servers into this product so it can use their tools G Agent access	ai-native user	Agenticness — how well agents can access and operate the productAgenticness	3	full	7/10	C claimed
Point an agent at llms.txt or agent-oriented docs G Agent access	ai-native user	Agenticness — how well agents can access and operate the productAgenticness	2	full	8/10	T probed
Subscribe to events via webhooks G Agent access	ai-native user	Agenticness — how well agents can access and operate the productAgenticness	2	full	8/10	C claimed
Build against official SDKs G Agent access	ai-native user	Agenticness — how well agents can access and operate the productAgenticness	2	partial	6/10	T probed
Get AI-generated insights and suggestions from my data inside the product G Agentic features	ai-native user	Agenticness — how well agents can access and operate the productAgenticness	2	partial	6/10	C claimed
Operate the product with natural-language commands G Agentic features	ai-native user	Agenticness — how well agents can access and operate the productAgenticness	2	partial	6/10	C claimed
Run the product headlessly / in CI for automation G Agent access	ai-native user	Agenticness — how well agents can access and operate the productAgenticness	2	partial	6/10	C claimed
Set up automations that run autonomously in the background G Agentic features	ai-native user	Agenticness — how well agents can access and operate the productAgenticness	2	partial	6/10	X community
Download a machine-readable API spec (OpenAPI or equivalent) G Api quality	ai-native user	Agenticness — how well agents can access and operate the productAgenticness	2	none	0/10
Explore an interactive API reference with runnable examples G Api quality	ai-native user	Agenticness — how well agents can access and operate the productAgenticness	2	none	0/10
Issue scoped/least-privilege API credentials for an agent G Agent access	ai-native user	Agenticness — how well agents can access and operate the productAgenticness	2	none	0/10
Rely on versioned APIs with a documented deprecation policy G Api quality	ai-native user	Agenticness — how well agents can access and operate the productAgenticness	2	none	0/10
Use an official CLI G Agent access	ai-native user	Agenticness — how well agents can access and operate the productAgenticness	2	none	untested		none yet
Test against a sandbox environment without touching production data G Api quality	ai-native user	Agenticness — how well agents can access and operate the productAgenticness	1	partial	4/10	C claimed
The agent ingests my help center, docs, past tickets, and internal wikis as knowledge sources without manual re-authoring C Ingestion	support ops lead	Knowledge grounding — stories about knowledge grounding in this arenaKnowledge grounding	3	full	8/10	C claimed
The agent runs inside my existing helpdesk — Zendesk, Salesforce, Intercom — or standalone, syncing tickets and context both ways C Helpdesk	developer	Integrations platform — stories about integrations platform in this arenaIntegrations platform	3	full	8/10	C claimed
The agent fully resolves a meaningful share of conversations end-to-end — measured as resolutions, not mere deflections or bounces C Resolution	support leader	Resolution quality — stories about resolution quality in this arenaResolution quality	3	full	7/10	X community
Define rules that trigger actions automatically on events G	ai-native user	Automation depth — how much of the product can run unattendedAutomation depth	3	partial	6/10	C claimed
Guardrails stop the agent from inventing policies, prices, or promises — off-knowledge questions get a safe decline, not a guess C Hallucination	ai-native user	Guardrails safety — stories about guardrails safety in this arenaGuardrails safety	3	partial	6/10	C claimed
The agent takes real actions through my APIs — refunds, order changes, subscription updates — with scoped auth per action C Actions	developer	Agent actions — stories about agent actions in this arenaAgent actions	3	partial	6/10	C claimed
When the agent escalates, the human gets the full conversation, a summary, and collected details — the customer never repeats themselves C Handoff	support leader	Escalation handoff — stories about escalation handoff in this arenaEscalation handoff	3	partial	6/10	C claimed
Every answer is grounded in my own content and shows which article or source it drew from C Grounding	ai-native user	Knowledge grounding — stories about knowledge grounding in this arenaKnowledge grounding	3	partial	5/10	C claimed
Dashboards show resolution rate, CSAT, handoff rate, and cost per resolution — the numbers I report to my exec team C Analytics	support leader	Insights analytics — stories about insights analytics in this arenaInsights analytics	3	partial	4/10	C claimed
Export all of my data in open formats and leave G	ai-native user	Openness — open source, data portability, and self-hosting storiesOpenness	3	partial	2/10	C claimed
Prevent my data from being used to train AI models G	ai-native user	Privacy posture — data-handling and privacy storiesPrivacy posture	3	none	untested		none yet
Self-host the core product G	ai-native user	Openness — open source, data portability, and self-hosting storiesOpenness	3	n/a	untested		none yet
I encode standard operating procedures the agent follows step-by-step for known issue types, with deterministic branching C Procedures	support ops lead	Agent actions — stories about agent actions in this arenaAgent actions	2	full	8/10	C claimed
Answers use the customer's live data — plan, order status, account history — not just generic help articles C Personalization	support leader	Resolution quality — stories about resolution quality in this arenaResolution quality	2	full	7/10	T probed
I test the agent against historical tickets or simulated conversations before it faces real customers C Simulation	support ops lead	Testing qa — stories about testing qa in this arenaTesting qa	2	full	7/10	C claimed
One agent covers chat, email, and in-app, plus the channels my customers actually use — Slack, WhatsApp, social C Channels	support leader	Channels languages — stories about channels languages in this arenaChannels languages	2	full	7/10	C claimed
The agent handles phone calls — speech in, speech out — with the same knowledge and actions as chat C Voice	support leader	Channels languages — stories about channels languages in this arenaChannels languages	2	full	7/10	C claimed
The agent supports customers in many languages, even where my knowledge base exists only in English C Languages	support leader	Channels languages — stories about channels languages in this arenaChannels languages	2	full	7/10	C claimed
I configure when the agent must hand off — by topic, sentiment, customer tier, or explicit request — and it reliably obeys C Rules	support ops lead	Escalation handoff — stories about escalation handoff in this arenaEscalation handoff	2	partial	6/10	C claimed
I mark topics as human-only — legal threats, cancellations, security — and the agent never freelances on them C Topic controls	support ops lead	Guardrails safety — stories about guardrails safety in this arenaGuardrails safety	2	partial	6/10	C claimed
The agent asks clarifying questions and works through multi-step troubleshooting instead of dumping one canned answer C Reasoning	support leader	Resolution quality — stories about resolution quality in this arenaResolution quality	2	partial	6/10	X community
Knowledge stays current automatically — the agent re-syncs sources on a schedule or on change, not via manual re-uploads C Freshness	support ops lead	Knowledge grounding — stories about knowledge grounding in this arenaKnowledge grounding	2	partial	5/10	C claimed
Do everything through the API that I can do in the UI G	ai-native user	Openness — open source, data portability, and self-hosting storiesOpenness	2	partial	4/10	T probed
Pricing is outcome-based and published — I pay per resolution with caps and controls, not an opaque enterprise quote G Pricing	support leader	Pricing economics — stories about pricing economics in this arenaPricing economics	2	partialpaid	4/10	X community
Perform bulk operations across many items at once G	ai-native user	Automation depth — how much of the product can run unattendedAutomation depth	2	partial	3/10	C claimed
Launch in a supervised mode where the agent drafts replies for human approval before anything reaches a customer C Supervision	support ops lead	Guardrails safety — stories about guardrails safety in this arenaGuardrails safety	2	none	0/10
Schedule recurring jobs or workflows G	ai-native user	Automation depth — how much of the product can run unattendedAutomation depth	2	none	0/10
Choose where my data is stored (region/residency) G	ai-native user	Privacy posture — data-handling and privacy storiesPrivacy posture	2	none	untested		none yet
Control data retention and deletion G	ai-native user	Privacy posture — data-handling and privacy storiesPrivacy posture	2	none	untested		none yet
Opt out of telemetry and usage tracking G	ai-native user	Privacy posture — data-handling and privacy storiesPrivacy posture	2	none	untested		none yet
Read the product's source under an open license G	ai-native user	Openness — open source, data portability, and self-hosting storiesOpenness	2	n/a	untested		none yet
AI conversations get ongoing QA — scored samples, flagged failures, and a review loop that feeds fixes back into the agent C Qa	support ops lead	Testing qa — stories about testing qa in this arenaTesting qa	1	full	7/10	C claimed
I control the agent's tone and brand voice, and it stays consistent across topics and languages C Voice	support leader	Resolution quality — stories about resolution quality in this arenaResolution quality	1	partial	6/10	C claimed
The platform clusters conversations by topic and surfaces emerging product issues before they spike ticket volume C Insights	support leader	Insights analytics — stories about insights analytics in this arenaInsights analytics	1	partial	6/10	C claimed
The platform surfaces knowledge gaps and conflicting content that cause the agent to miss or fumble questions C Gaps	support ops lead	Knowledge grounding — stories about knowledge grounding in this arenaKnowledge grounding	1	partial	4/10	C claimed
Version, review, and roll back my automations G	ai-native user	Automation depth — how much of the product can run unattendedAutomation depth	1	none	0/10

Opportunities — the stories that would move this product's scores, from its own judged verdictsOpportunitiestop 8 of 36 stories with headroom

What would move Fin’s scores — derived from its own judged verdicts, biggest headroom first. Each line quotes what the judge found missing; shipping it (or evidencing it publicly) is the fix.

Privacy posture — data-handling and privacy storiesPrevent my data from being used to train AI models
nonemoves PA Scoreimpact 30
No evidence in the pack addresses data being used for AI model training or an opt-out/data-use control; general trust/compliance mentions (e.g., trust-reliability page) do not specify training data usage or opt-out mechanisms.
Agenticness — how well agents can access and operate the productUse an official CLI
nonemoves agent-readyimpact 30
Evidence shows Fin has an API, Node SDK, and MCP server, but no official CLI tool is documented or referenced anywhere in the evidence pack.
Agenticness — how well agents can access and operate the productIssue scoped/least-privilege API credentials for an agent
nonemoves agent-readyimpact 30
Missing: scoped/least-privilege credential issuance, API key/token permission granularity, agent-specific credential management docs.
Agenticness — how well agents can access and operate the productExplore an interactive API reference with runnable examples
nonemoves API qualityimpact 30
While Fin has API docs (developers.intercom.com) and an SDK on GitHub, there is no evidence of an interactive API reference with runnable/try-it examples; a probe explicitly found no OpenAPI/Swagger spec published at any candidate path, and no docs mention a live API console.
Agenticness — how well agents can access and operate the productDownload a machine-readable API spec (OpenAPI or equivalent)
nonemoves API qualityimpact 30
Fin has documented REST APIs (Fin Agent API, Node/TS SDK) but no evidence of a downloadable machine-readable spec; a direct probe for OpenAPI/swagger files at common paths returned 404 for all candidates, and no docs page links such a spec.
Agenticness — how well agents can access and operate the productRely on versioned APIs with a documented deprecation policy
nonemoves API qualityimpact 30
The evidence shows Fin has a REST API with a 'Preview API version' for new endpoints (docs-4) and a general changes/changelog page (docs-1), but there is no documented versioning scheme or explicit deprecation policy (e.g., version sunset timelines, backward-compatibility guarantees) cited anywhere in the pack.
Openness — open source, data portability, and self-hosting storiesExport all of my data in open formats and leave
partialq2/10moves PA Scoreimpact 24
Missing: bulk/full account data export, documented open file formats (CSV/JSON), a migration-out or offboarding process, and any independent confirmation of successful full data extraction.
Automation depth — how much of the product can run unattendedSchedule recurring jobs or workflows
nonemoves PA Scoreimpact 20
Missing: any documentation of scheduled/recurring workflow triggers, cron-style job scheduling, or recurring automation configuration.

Showing the top 8 of 36 — every none/partial verdict in the story verdicts table is headroom.

Think a verdict is wrong? Every verdicts-table row has a Flag link — see the methodology.

Coverage map — which docs area, API section, or community source covers which judged storiesCoverage map20 surfaces · 39 covered stories

Where the cited evidence behind each covered verdict came from — the same citations the verdicts table shows, no extra judging.

Help docs24 stories

The agent takes real actions through my APIs — refunds, order changes, subscription updates — with scoped auth per action
I encode standard operating procedures the agent follows step-by-step for known issue types, with deterministic branching
Point an agent at llms.txt or agent-oriented docs
Plug MCP servers into this product so it can use their tools
Set up automations that run autonomously in the background
Delegate tasks to a built-in AI assistant inside the product
Operate the product with natural-language commands
Test against a sandbox environment without touching production data
Define rules that trigger actions automatically on events
The agent supports customers in many languages, even where my knowledge base exists only in English
The agent handles phone calls — speech in, speech out — with the same knowledge and actions as chat
When the agent escalates, the human gets the full conversation, a summary, and collected details — the customer never repeats themselves
I configure when the agent must hand off — by topic, sentiment, customer tier, or explicit request — and it reliably obeys
Guardrails stop the agent from inventing policies, prices, or promises — off-knowledge questions get a safe decline, not a guess
I mark topics as human-only — legal threats, cancellations, security — and the agent never freelances on them
The agent runs inside my existing helpdesk — Zendesk, Salesforce, Intercom — or standalone, syncing tickets and context both ways
Knowledge stays current automatically — the agent re-syncs sources on a schedule or on change, not via manual re-uploads
The platform surfaces knowledge gaps and conflicting content that cause the agent to miss or fumble questions
Every answer is grounded in my own content and shows which article or source it drew from
The agent ingests my help center, docs, past tickets, and internal wikis as knowledge sources without manual re-authoring
The agent asks clarifying questions and works through multi-step troubleshooting instead of dumping one canned answer
I control the agent's tone and brand voice, and it stays consistent across topics and languages
AI conversations get ongoing QA — scored samples, flagged failures, and a review loop that feeds fixes back into the agent
I test the agent against historical tickets or simulated conversations before it faces real customers

docs16 stories

Point an agent at llms.txt or agent-oriented docs
Run the product headlessly / in CI for automation
Plug MCP servers into this product so it can use their tools
Connect an agent via an official MCP server
Drive the product through a documented public API
Build against official SDKs
Subscribe to events via webhooks
Delegate tasks to a built-in AI assistant inside the product
Perform bulk operations across many items at once
Define rules that trigger actions automatically on events
When the agent escalates, the human gets the full conversation, a summary, and collected details — the customer never repeats themselves
I configure when the agent must hand off — by topic, sentiment, customer tier, or explicit request — and it reliably obeys
I mark topics as human-only — legal threats, cancellations, security — and the agent never freelances on them
The agent runs inside my existing helpdesk — Zendesk, Salesforce, Intercom — or standalone, syncing tickets and context both ways
Do everything through the API that I can do in the UI
Answers use the customer's live data — plan, order status, account history — not just generic help articles

Pricing docs16 stories

The agent takes real actions through my APIs — refunds, order changes, subscription updates — with scoped auth per action
Get AI-generated insights and suggestions from my data inside the product
Set up automations that run autonomously in the background
Delegate tasks to a built-in AI assistant inside the product
Define rules that trigger actions automatically on events
One agent covers chat, email, and in-app, plus the channels my customers actually use — Slack, WhatsApp, social
When the agent escalates, the human gets the full conversation, a summary, and collected details — the customer never repeats themselves
I configure when the agent must hand off — by topic, sentiment, customer tier, or explicit request — and it reliably obeys
Dashboards show resolution rate, CSAT, handoff rate, and cost per resolution — the numbers I report to my exec team
The platform clusters conversations by topic and surfaces emerging product issues before they spike ticket volume
The agent runs inside my existing helpdesk — Zendesk, Salesforce, Intercom — or standalone, syncing tickets and context both ways
The platform surfaces knowledge gaps and conflicting content that cause the agent to miss or fumble questions
Pricing is outcome-based and published — I pay per resolution with caps and controls, not an opaque enterprise quote
Answers use the customer's live data — plan, order status, account history — not just generic help articles
I control the agent's tone and brand voice, and it stays consistent across topics and languages
AI conversations get ongoing QA — scored samples, flagged failures, and a review loop that feeds fixes back into the agent

Integrations docs9 stories

Plug MCP servers into this product so it can use their tools
Connect an agent via an official MCP server
Set up automations that run autonomously in the background
Delegate tasks to a built-in AI assistant inside the product
Dashboards show resolution rate, CSAT, handoff rate, and cost per resolution — the numbers I report to my exec team
The agent runs inside my existing helpdesk — Zendesk, Salesforce, Intercom — or standalone, syncing tickets and context both ways
The agent ingests my help center, docs, past tickets, and internal wikis as knowledge sources without manual re-authoring
The agent asks clarifying questions and works through multi-step troubleshooting instead of dumping one canned answer
The agent fully resolves a meaningful share of conversations end-to-end — measured as resolutions, not mere deflections or bounces

GitHub README8 stories

Run the product headlessly / in CI for automation
Drive the product through a documented public API
Build against official SDKs
Perform bulk operations across many items at once
Knowledge stays current automatically — the agent re-syncs sources on a schedule or on change, not via manual re-uploads
Every answer is grounded in my own content and shows which article or source it drew from
The agent ingests my help center, docs, past tickets, and internal wikis as knowledge sources without manual re-authoring
Do everything through the API that I can do in the UI

Procedures docs7 stories

The agent takes real actions through my APIs — refunds, order changes, subscription updates — with scoped auth per action
I encode standard operating procedures the agent follows step-by-step for known issue types, with deterministic branching
Set up automations that run autonomously in the background
Delegate tasks to a built-in AI assistant inside the product
Operate the product with natural-language commands
Answers use the customer's live data — plan, order status, account history — not just generic help articles
The agent asks clarifying questions and works through multi-step troubleshooting instead of dumping one canned answer

Analyze docs6 stories

Get AI-generated insights and suggestions from my data inside the product
Dashboards show resolution rate, CSAT, handoff rate, and cost per resolution — the numbers I report to my exec team
The platform clusters conversations by topic and surfaces emerging product issues before they spike ticket volume
The platform surfaces knowledge gaps and conflicting content that cause the agent to miss or fumble questions
The agent fully resolves a meaningful share of conversations end-to-end — measured as resolutions, not mere deflections or bounces
AI conversations get ongoing QA — scored samples, flagged failures, and a review loop that feeds fixes back into the agent

Testing docs6 stories

I encode standard operating procedures the agent follows step-by-step for known issue types, with deterministic branching
Test against a sandbox environment without touching production data
Guardrails stop the agent from inventing policies, prices, or promises — off-knowledge questions get a safe decline, not a guess
The agent fully resolves a meaningful share of conversations end-to-end — measured as resolutions, not mere deflections or bounces
AI conversations get ongoing QA — scored samples, flagged failures, and a review loop that feeds fixes back into the agent
I test the agent against historical tickets or simulated conversations before it faces real customers

Hacker News5 stories

Set up automations that run autonomously in the background
Delegate tasks to a built-in AI assistant inside the product
Pricing is outcome-based and published — I pay per resolution with caps and controls, not an opaque enterprise quote
The agent asks clarifying questions and works through multi-step troubleshooting instead of dumping one canned answer
The agent fully resolves a meaningful share of conversations end-to-end — measured as resolutions, not mere deflections or bounces

API platform docs5 stories

The agent takes real actions through my APIs — refunds, order changes, subscription updates — with scoped auth per action
Drive the product through a documented public API
Build against official SDKs
Every answer is grounded in my own content and shows which article or source it drew from
Do everything through the API that I can do in the UI

OpenAPI spec4 stories

Point an agent at llms.txt or agent-oriented docs
Drive the product through a documented public API
Build against official SDKs
Do everything through the API that I can do in the UI

Integrations docs4 stories

The agent runs inside my existing helpdesk — Zendesk, Salesforce, Intercom — or standalone, syncing tickets and context both ways
The agent ingests my help center, docs, past tickets, and internal wikis as knowledge sources without manual re-authoring
Export all of my data in open formats and leave
Answers use the customer's live data — plan, order status, account history — not just generic help articles

Train docs4 stories

Operate the product with natural-language commands
Guardrails stop the agent from inventing policies, prices, or promises — off-knowledge questions get a safe decline, not a guess
Every answer is grounded in my own content and shows which article or source it drew from
I control the agent's tone and brand voice, and it stays consistent across topics and languages

Capabilities docs3 stories

I encode standard operating procedures the agent follows step-by-step for known issue types, with deterministic branching
Get AI-generated insights and suggestions from my data inside the product
I test the agent against historical tickets or simulated conversations before it faces real customers

Changes docs3 stories

The agent takes real actions through my APIs — refunds, order changes, subscription updates — with scoped auth per action
Export all of my data in open formats and leave
Answers use the customer's live data — plan, order status, account history — not just generic help articles

Channels docs3 stories

One agent covers chat, email, and in-app, plus the channels my customers actually use — Slack, WhatsApp, social
The agent handles phone calls — speech in, speech out — with the same knowledge and actions as chat
I control the agent's tone and brand voice, and it stays consistent across topics and languages

Voice docs3 stories

The agent handles phone calls — speech in, speech out — with the same knowledge and actions as chat
Guardrails stop the agent from inventing policies, prices, or promises — off-knowledge questions get a safe decline, not a guess
Every answer is grounded in my own content and shows which article or source it drew from

CLI docs2 stories

Point an agent at llms.txt or agent-oriented docs
Operate the product with natural-language commands

Trust reliability docs2 stories

The agent takes real actions through my APIs — refunds, order changes, subscription updates — with scoped auth per action
Guardrails stop the agent from inventing policies, prices, or promises — off-knowledge questions get a safe decline, not a guess

llms.txt1 story

Point an agent at llms.txt or agent-oriented docs

Probe proofs — replayable recordings from the probe harnessProbe proofs

Replayable recordings from our probe harness — see the Prove-It protocol to submit one.

$curl -si https://api.intercom.io/mereproduced

$ curl -si https://api.intercom.io/me
HTTP/2 401

date: Thu, 10 Sep 2026 18:40:31 GMT

content-type: application/json; charset=utf-8

status: 401 Unauthorized

vary: Accept

x-intercom-version: ab29b1c2ca8eb625f435dc65192eb53e4b939455

www-authenticate: Basic realm="intercom.io"

x-request-id: 0008s9g5tlv28u1uodig

x-frame-options: SAMEORIGIN

cache-control: no-cache

strict-transport-security: max-age=31556952; includeSubDomains; preload

referrer-policy: strict-origin-when-cross-origin

x-xss-protection: 1; mode=block

x-request-queueing: 0

x-runtime: 0.005443

x-content-type-options: nosniff

server: nginx

{"type":"error.list","request_id":"0008s9g5tlv28u1uodig","errors":[{"code":"missing_authorization","message":"No authorization was provided"}]}

proves: Drive the product through a documented public API proves: Issue scoped/least-privilege API credentials for an agentrecorded 2026-09-10

$curl -s https://developers.intercom.com/llms.txt | head -6reproduced

$ curl -s https://developers.intercom.com/llms.txt | head -6
# Intercom and Fin Developer Platform

> Faster resolutions, higher CSAT, and lighter support volumes with the only platform to combine the power of automation and human customer support.

## Table of contents

proves: Point an agent at llms.txt or agent-oriented docsrecorded 2026-09-10

$curl -si -X POST https://mcp.intercom.com/mcp -H 'Content-Type: application/json' -d '<jsonrpc initialize>'reproduced

$ curl -si -X POST https://mcp.intercom.com/mcp -H 'Content-Type: application/json' -d '<jsonrpc initialize>'
HTTP/2 401

date: Thu, 10 Sep 2026 18:40:31 GMT

content-type: application/json

content-length: 79

www-authenticate: Bearer realm="OAuth", error="invalid_[redacted]", error_description="Missing or invalid access [redacted]"

set-cookie: __cf_bm=VJQgto64veZeBL_slnJThB.EzkI4GGs7jSUR_xLKrm0-1789065631.7702801-1.0.1.1-ZxNW0IL8QLABdSDiuHPkWx_xMLh6TWqBQd47.sFAZOLby8UHKgtrimNSSBysFecQ8P.hpA2m0vBetCjwlxXGIvN5CYnz6ppiHP4MIdVYyYMBDBARE4aYuz85HkiDYudV; HttpOnly; SameSite=None; Secure; Path=/; Domain=mcp.intercom.com; Expires=Thu, 10 Sep 2026 19:10:31 GMT

server: cloudflare

cf-ray: a3908fc68d1eb917-SJC

{"error":"invalid_[redacted]","error_description":"Missing or invalid access [redacted]"}

proves: Connect an agent via an official MCP serverrecorded 2026-09-10

$curl -s https://fin.ai/llms.txt | head -6reproduced

$ curl -s https://fin.ai/llms.txt | head -6
# Fin — llms.txt

> This file helps large language models (LLMs) understand the content structure, canonical sources, and retrieval rules for Fin.ai and its subdomains.

---

proves: Point an agent at llms.txt or agent-oriented docsrecorded 2026-09-10

$npm view intercom-client name versionreproduced

$ npm view intercom-client name version
name = 'intercom-client'
version = '7.0.3'

proves: Build against official SDKsrecorded 2026-09-10

Claims vs evidence — vendor claims reconciled against independent verdictsClaims vs evidence

4 of 19 testable claims verified · 0 contradicted → integrity 21/100

27 distinct capability claims found in Fin’s own claimed-docs/GitHub materials, reconciled against our judge’s independent verdicts.

Verified

Unverified

Contradicted

Undersold

Verified (4)

“Fin can be accessed and controlled programmatically via an API”
Drive the product through a documented public API partial proof ↗
“Supports Model Context Protocol so AI agents can securely access and interact with Intercom data”
Connect an agent via an official MCP server full proof ↗
“Official SDK provides iterators for paginated list endpoints”
Build against official SDKs partial proof ↗
“Fin supports 'Tasks' for executing defined actions/workflows on behalf of users”
Set up automations that run autonomously in the background partial proof ↗

Unverified (19)

“Dedicated orchestration API endpoints let you discover capabilities, ask Fin, run a procedure, or escalate to a human”
The agent takes real actions through my APIs — refunds, order changes, subscription updates — with scoped auth per action partial proof ↗
“Webhooks let you subscribe to real-time events like new contacts, conversations, or outbound message receipts”
Subscribe to events via webhooks full proof ↗
“Fin AI Agent works with any helpdesk, including Salesforce and HubSpot, not just Intercom's own inbox”
The agent runs inside my existing helpdesk — Zendesk, Salesforce, Intercom — or standalone, syncing tickets and context both ways full proof ↗
“Handles tickets, cases, emails, live chat, WhatsApp, SMS and more across channels”
One agent covers chat, email, and in-app, plus the channels my customers actually use — Slack, WhatsApp, social full proof ↗
“Tone and answer length can be customized”
I control the agent's tone and brand voice, and it stays consistent across topics and languages partial proof ↗
“Fin can take actions that update external systems, not just answer questions”
The agent takes real actions through my APIs — refunds, order changes, subscription updates — with scoped auth per action partial proof ↗
“Escalations transfer directly into the support team's preferred inbox”
When the agent escalates, the human gets the full conversation, a summary, and collected details — the customer never repeats themselves partial proof ↗
“Includes a public Help Center and Knowledge Hub as content sources”
The agent ingests my help center, docs, past tickets, and internal wikis as knowledge sources without manual re-authoring full proof ↗
“Provides analytics like CX Score, AI Topics, Trends, and AI Recommendations for surfacing patterns”
The platform clusters conversations by topic and surfaces emerging product issues before they spike ticket volume partial proof ↗
“SDK/API lets you import external website content as a knowledge source”
The agent ingests my help center, docs, past tickets, and internal wikis as knowledge sources without manual re-authoring full proof ↗
“Fin can be deployed to handle voice/phone calls”
The agent handles phone calls — speech in, speech out — with the same knowledge and actions as chat full proof ↗
“Fin AI Agent can respond to customers in multiple languages”
The agent supports customers in many languages, even where my knowledge base exists only in English full proof ↗
“You can give Fin explicit guidance to shape or restrict how it answers certain topics”
I mark topics as human-only — legal threats, cancellations, security — and the agent never freelances on them partial proof ↗
“Support teams can build step-by-step Procedures for Fin to follow on known issue types”
I encode standard operating procedures the agent follows step-by-step for known issue types, with deterministic branching full proof ↗
“Escalation guidance and rules can be configured to control when Fin hands off to a human”
I configure when the agent must hand off — by topic, sentiment, customer tier, or explicit request — and it reliably obeys partial proof ↗
“Websites can be synced and kept up to date automatically as a knowledge source”
Knowledge stays current automatically — the agent re-syncs sources on a schedule or on change, not via manual re-uploads partial proof ↗
“Content can be added to Fin's knowledge base from existing sources”
The agent ingests my help center, docs, past tickets, and internal wikis as knowledge sources without manual re-authoring full proof ↗
“Fin can be previewed and tested before going live with real customers”
I test the agent against historical tickets or simulated conversations before it faces real customers full proof ↗
“Provides a JavaScript-based mechanism to hand off conversations to a human”
When the agent escalates, the human gets the full conversation, a summary, and collected details — the customer never repeats themselves partial proof ↗

Undersold (20)

Point an agent at llms.txt or agent-oriented docs full proof ↗
Run the product headlessly / in CI for automation partial proof ↗
Plug MCP servers into this product so it can use their tools full proof ↗
Get AI-generated insights and suggestions from my data inside the product partial proof ↗
Delegate tasks to a built-in AI assistant inside the product full proof ↗
Operate the product with natural-language commands partial proof ↗
Test against a sandbox environment without touching production data partial proof ↗
Perform bulk operations across many items at once partial proof ↗
Define rules that trigger actions automatically on events partial proof ↗
Guardrails stop the agent from inventing policies, prices, or promises — off-knowledge questions get a safe decline, not a guess partial proof ↗
Dashboards show resolution rate, CSAT, handoff rate, and cost per resolution — the numbers I report to my exec team partial proof ↗
The platform surfaces knowledge gaps and conflicting content that cause the agent to miss or fumble questions partial proof ↗
Every answer is grounded in my own content and shows which article or source it drew from partial proof ↗
Do everything through the API that I can do in the UI partial proof ↗
Export all of my data in open formats and leave partial proof ↗
Pricing is outcome-based and published — I pay per resolution with caps and controls, not an opaque enterprise quote partial proof ↗
Answers use the customer's live data — plan, order status, account history — not just generic help articles full proof ↗
The agent asks clarifying questions and works through multi-step troubleshooting instead of dumping one canned answer partial proof ↗
The agent fully resolves a meaningful share of conversations end-to-end — measured as resolutions, not mere deflections or bounces full proof ↗
AI conversations get ongoing QA — scored samples, flagged failures, and a review loop that feeds fixes back into the agent full proof ↗

Claims outside our story set (4)

Real capability claims found in Fin’s own materials, but no story in this arena’s taxonomy covers them yet — that’s feedback on the taxonomy, not a mark against the product.

“Includes a suite of proactive outbound messaging tools”
source ↗
“Lets you inspect exact error timing and status codes and confirm a fix worked”
source ↗
“You can export a saved View exactly as displayed”
source ↗
“Includes disclosure functionality to inform users they are talking to an AI agent”
source ↗

Suggest a story for these →

Business model

usage-basedsubscription-per-seatenterprise-custom

$0.99 per Fin outcome/resolution ($9.99 per qualification outcome), 50-outcome monthly minimum; optional helpdesk seats $29/seat/mo, Copilot $35/user/mo; Fin Voice custom-priced; 14-day free trial.

pricing ↗

Score trend

How this product’s scores have moved as evidence and verdicts are re-derived — a point per change, not per day.

PA Score34 (Sep 10 '26)29 (Sep 16 '26)

Agent-ready56 (Sep 10 '26)50 (Sep 16 '26)

Try Experimental

Run it in the microterminal →

Recorded agent sessions — and a live MCP handshake where the vendor ships one.

Flag

⚑ Flag a verdict

Think a verdict is wrong? Opens a prefilled GitHub issue — or use the ⚑ next to any verdict above.

Badge

Embed this product's score badge →

Hotlinked SVG — always shows the live current score.

For agents

This page as markdown /llms.txt

Data

Evidence (JSON)Verdicts (JSON)

Agent surface uptime MCP up · llms.txt up (tracking since Sep 11 '26)

Fin

Products

Try itExperimental

Verified integrations

By theme — the product's score on each story themeBy theme

Story verdicts — every judged story with its evidenceStory verdicts

🤖Agent actions — stories about agent actions in this arenaAgent actions

🤖Agenticness — how well agents can access and operate the productAgenticness

⚙️Automation depth — how much of the product can run unattendedAutomation depth

💬Channels languages — stories about channels languages in this arenaChannels languages

🚨Escalation handoff — stories about escalation handoff in this arenaEscalation handoff

🛡️Guardrails safety — stories about guardrails safety in this arenaGuardrails safety

📊Insights analytics — stories about insights analytics in this arenaInsights analytics

🔌Integrations platform — stories about integrations platform in this arenaIntegrations platform

📚Knowledge grounding — stories about knowledge grounding in this arenaKnowledge grounding

🔓Openness — open source, data portability, and self-hosting storiesOpenness

💲Pricing economics — stories about pricing economics in this arenaPricing economics

🔒Privacy posture — data-handling and privacy storiesPrivacy posture

🧪Resolution quality — stories about resolution quality in this arenaResolution quality

🧪Testing qa — stories about testing qa in this arenaTesting qa

Opportunities — the stories that would move this product's scores, from its own judged verdictsOpportunitiestop 8 of 36 stories with headroom

Coverage map — which docs area, API section, or community source covers which judged storiesCoverage map20 surfaces · 39 covered stories

Probe proofs — replayable recordings from the probe harnessProbe proofs

Claims vs evidence — vendor claims reconciled against independent verdictsClaims vs evidence

Business model

Score trend

Agent actions — stories about agent actions in this arenaAgent actions

Agenticness — how well agents can access and operate the productAgenticness

Automation depth — how much of the product can run unattendedAutomation depth

Channels languages — stories about channels languages in this arenaChannels languages

Escalation handoff — stories about escalation handoff in this arenaEscalation handoff

Guardrails safety — stories about guardrails safety in this arenaGuardrails safety

Insights analytics — stories about insights analytics in this arenaInsights analytics

Integrations platform — stories about integrations platform in this arenaIntegrations platform

Knowledge grounding — stories about knowledge grounding in this arenaKnowledge grounding

Openness — open source, data portability, and self-hosting storiesOpenness

Pricing economics — stories about pricing economics in this arenaPricing economics

Privacy posture — data-handling and privacy storiesPrivacy posture

Resolution quality — stories about resolution quality in this arenaResolution quality

Testing qa — stories about testing qa in this arenaTesting qa