customer-facing support, conversation data
Kill Intercom
$74-$400+/mo
Intercom is powerful but its per-resolution AI pricing ($0.99 per AI-resolved conversation) and base seat fees make it expensive and unpredictable for small teams. Chatwoot is open source, self-hostable, covers live chat, ticketing, and knowledge base - everything most small teams actually need. Layer a Claude-powered AI auto-responder on top for a fraction of Intercom's Fin AI cost.
Why Replace Intercom?
- ×Intercom's Fin AI charges $0.99 per resolved conversation - a few hundred support chats and you're paying more than the base plan
- ×The Starter plan is $74/mo but most useful features (custom bots, product tours, A/B testing) require the $153+/mo plans
- ×Most small teams use Intercom for live chat and a basic help center - you don't need a $400/mo platform for that
- ×Self-hosting Chatwoot gives you full control over your support data and zero per-conversation fees for AI responses
What People Use Intercom For
Replacement Stack
Chatwoot
Open source customer support platform with live chat, email integration, ticketing, canned responses, and a knowledge base. Self-hosted means you own all your support data. The UI is clean and your team can be up and running in an afternoon.
Railway
One-click hosting for Chatwoot. Railway handles the server, database, Redis, and Sidekiq workers that Chatwoot needs. The $5/mo Hobby plan is enough for most small teams. Scales up automatically if you get traffic spikes.
Vercel
Free hosting for your AI auto-responder webhook. When a customer sends a message, Chatwoot fires a webhook to your Vercel function, which calls Claude and posts the AI response back. Zero cost at low volume.
Supabase
Optional but useful - store your FAQ content, conversation logs, and AI response analytics in a Supabase database. The free tier handles most small teams. Also useful if you want to build a custom knowledge base that feeds your AI responder.
Anthropic API
Powers your AI auto-responder. Feed it your FAQ content as context and it handles common questions automatically. Haiku model costs ~$0.001 per response vs Intercom's $0.99 per resolution - that's roughly 1,000x cheaper per conversation.
Kill Checklist
Export your Intercom data
~15 minsLog into Intercom > Settings > Data Management > Export your data. Download your conversation history, contact list, help center articles, and saved replies. The help center articles are especially important - you'll use these as the knowledge base for your Claude AI responder. Export help articles as HTML or markdown if possible.
Sign up for Railway and deploy Chatwoot
~15 minsGo to railway.app and create an account (GitHub login works). Search the Railway template marketplace for 'Chatwoot' - there's an official one-click template. Click Deploy, wait 3-5 minutes for it to spin up. Railway automatically provisions the PostgreSQL database and Redis instance that Chatwoot needs. Once deployed, you'll get a URL like chatwoot-production-xxxx.up.railway.app - this is your support dashboard.
https://railway.com?referralCode=killyoursaas →Configure Chatwoot basics
~30 minsOpen your Chatwoot dashboard and set up: (1) Create your account and first agent, (2) Go to Settings > Inboxes > Add Inbox > Website - this generates your chat widget code, (3) Add canned responses for your most common replies (Settings > Canned Responses), (4) Create labels for ticket categorization (Settings > Labels), (5) Set up your team hours and auto-reply messages. Import your Intercom saved replies as canned responses.
Get the full kill plan for Intercom
Sign up to unlock 7 more steps, a customizable Claude Code prompt tailored to the features you actually use, and a checklist to track your progress.
- ✓Pick which Intercom features you care about - skip the rest
- ✓Get a Claude prompt customized to your exact setup
- ✓Track your kill progress across all your SaaS tools
Free account. No credit card required.