Intercom Fin alternative: a support agent built with Claude Code
A software company replaced Intercom's Fin with its own AI agent, built with Claude Code. It answers every customer conversation and reads screenshots. It checks Linear to see whether a bug is fixed, links the right Canny roadmap item and refreshes pricing daily from the public page. It writes in the voice of 500 of the team's best past replies. Agents silence it on a conversation with @ai stop, and a categorized digest lands in Slack at 8am.
Setup steps
- Start with a one-week prototype that drafts replies in your help desk
- Teach it your voice from your best past replies (they used 500) and your docs
- Connect your bug tracker, roadmap and a daily pricing refresh
- Add a safety pipeline: no links to help articles that don't exist, and risky drafts posted as internal notes
- Build an eval suite that grades drafts on accuracy, tone and policy before any change ships
- Add @ai stop and @ai start commands and a daily Slack digest
Steps summarized from the original Blog source ↗. Use “Copy” to grab them with the tools and source link, or copy an AI prompt.
How it went
Three months and 44 deploys after a one-week experiment, the agent handles every customer conversation for about $50 a month, against about $800 for Fin. The company left Fin after it sent a customer to a help article that didn't exist and quoted a plan price in pounds when the company sells in dollars.
Worth knowing
An MVP is not a product: the first version worked because volume was low. Rules piled into a system prompt start silently breaking earlier ones by about the twentieth, so an eval suite that catches regressions is what made it safe to leave running.
Only what the original source ↗ says, re-read Nov 2, 2026. How we write these
Try it with AI
Paste this into ChatGPT, Claude or your own agent for step-by-step help. Prompt written by AppToBot from this recipe's data, not by the original author. It keeps drafts for your review and asks before anything is sent or paid.
Help me set up this automation using In-house AI support agent built with Claude Code (Claude via AWS in the EU) + Linear + Canny + Slack: "Intercom Fin alternative: a support agent built with Claude Code". Goal: replace Intercom Fin AI support agent. How it should work: - Trigger: A customer writes in - Then: Reads the message and any screenshot - Then: Checks Linear, Canny and today's pricing - Then: A safety check runs; risky drafts become internal notes - Result: Reply sent, or a draft left for an agent Setup steps (summarized from the original source): 1. Start with a one-week prototype that drafts replies in your help desk 2. Teach it your voice from your best past replies (they used 500) and your docs 3. Connect your bug tracker, roadmap and a daily pricing refresh 4. Add a safety pipeline: no links to help articles that don't exist, and risky drafts posted as internal notes 5. Build an eval suite that grades drafts on accuracy, tone and policy before any change ships 6. Add @ai stop and @ai start commands and a daily Slack digest Guide me one step at a time. Ask me for anything you need (accounts, sheet or channel names) before you start. Safety rules: - Build and test with sample data first, and show me a sample result before anything goes live. - Keep anything that goes to other people (emails, posts, replies, record updates) as drafts for my review until I say otherwise. - Never send, post, buy, pay, delete or change records without my approval. - Tell me about any cost (paid plans, API usage) before I'd incur it, and help me set a spending limit. - Keep API keys and passwords out of shared docs and chats. - Privacy: Claude calls go through AWS in the EU for data residency, sub-processors are listed publicly, and one API call scrubs a customer's data on request (per company). Based on: https://marker.io/blog/we-built-our-own-ai-support-agent
Tool names link to each official site (opens in a new tab).
Running cost: API pennies Runs on the company's own code; it reports about $50 a month and names no subscription, but gives no breakdown. Self-hosted How we sort costs
Stack: Claude · Slack
Replace an app: AI bots that replace Intercom
App vs bot, side by side
| Old app / habit | Bot | |
|---|---|---|
| Tool | Intercom Fin | In-house support agent with a safety pipeline and an eval suite |
| Monthly cost | About $800/mo for Fin (per company) | About $50/mo (per company) |
| Setup | None (already in use) | Not reported · Advanced |
| Your time | Baseline | Not reported |
| Maintenance | Handled by vendor/you as before | Every behaviour change runs through the eval suite before it ships; 44 deploys in three months. |
| Privacy | Data stays with the app vendor | Claude calls go through AWS in the EU for data residency, sub-processors are listed publicly, and one API call scrubs a customer's data on request (per company). |
Full comparison
| Replaces | Intercom Fin |
|---|---|
| Bot / agent stack | In-house support agent with a safety pipeline and an eval suite |
| Monthly cost before | About $800/mo for Fin (per company) |
| Monthly cost after | About $50/mo (per company) |
| Setup time | Not reported |
| Time saved | Not reported |
| Skill level | Advanced |
| Reliability | 4/5: three months in production on every conversation, with human review built in (risky drafts become internal notes) and an eval suite gating changes. One company, and no accuracy figures are published. |
| Maintenance | Every behaviour change runs through the eval suite before it ships; 44 deploys in three months. |
| Privacy & data | Claude calls go through AWS in the EU for data residency, sub-processors are listed publicly, and one API call scrubs a customer's data on request (per company). |
| Best for | Small support teams with real integrations and an opinionated tone |
| Not for | A five-person team without those needs; the company says Fin is fine there |
Pros and cons
Pros
- About $750/mo saved (per company)
- Knows bug status, roadmap and current pricing
- Agents stop, restart or redirect it with one comment
Cons
- A real engineering project: the safety, eval and staging work took weeks
- You own every mistake and every fix
Source and evidence
Company blog post (2026) by its product lead: three months and 44 deploys in production, handling every customer conversation. A safety check runs before each reply, humans can take over, and an eval suite gates every change.
Read the original Blog source ↗
“Not reported” means the source doesn't say. We never fill gaps with guesses. Numbers marked “per author” or “vendor claim” are the source's own and weren't independently tested. Source last checked Nov 2, 2026.
Our rating is an editorial opinion based on the public source above, not lab testing. Spot an error? Request a correction.
Plain-text summary (for search and AI assistants)
Intercom Fin alternative: a support agent built with Claude Code. Replaces Intercom Fin with In-house support agent with a safety pipeline and an eval suite. Cost before: About $800/mo for Fin (per company). Cost after: About $50/mo (per company). Setup: Not reported, Advanced. Time saved: Not reported. Reliability 4/5: 4/5: three months in production on every conversation, with human review built in (risky drafts become internal notes) and an eval suite gating changes. One company, and no accuracy figures are published. Best for: Small support teams with real integrations and an opinionated tone. Not for: A five-person team without those needs; the company says Fin is fine there. How it went (per source): Three months and 44 deploys after a one-week experiment, the agent handles every customer conversation for about $50 a month, against about $800 for Fin. The company left Fin after it sent a customer to a help article that didn't exist and quoted a plan price in pounds when the company sells in dollars. Worth knowing: An MVP is not a product: the first version worked because volume was low. Rules piled into a system prompt start silently breaking earlier ones by about the twentieth, so an eval suite that catches regressions is what made it safe to leave running. Source: https://marker.io/blog/we-built-our-own-ai-support-agent
Related recipes
- AI agents that replaced 3 part-time contractors (n8n + Claude) (4/5)
- n8n customer support email triage with AI reply drafts (4/5)
- AI support ticket router with billing, technical and account agents (n8n) (3/5)
- Jira AI triage for support tickets, with suggested fixes (n8n) (3/5)
Questions, corrections or ideas?
Spotted an error, want a recipe covered, or have a partnership idea? Send us a message and we'll get back to you.
Starting your own newsletter? Try beehiiv ↗
What is beehiiv?
beehiiv is a newsletter platform: you write and send emails, host a newsletter website, and grow and monetize a list from one place. We've used it for our own newsletter.
beehiiv says (vendor claims, not tested by us):
- Free plan with up to 2,500 subscribers and unlimited sends
- Built-in website, recommendation network and analytics
- 0% cut of paid subscription revenue (Stripe fees apply)
Good for: starting or growing an email newsletter. Not ideal for: email automations or removing beehiiv branding on a $0 budget (those need Lite or Pro).
Free plan: $0, no card needed. Paid plans from $49/mo billed annually (Lite, up to 1,000 subscribers). People who sign up through our link get a 14-day trial plus 20% off their first 3 months, per beehiiv's partner page. Facts as of Oct 2026, from beehiiv's own website. More about beehiiv →