70% of your support tickets answered before you read them.
The rest escalated with full context. Confidence-scored. Fully auditable.
$99 one-time. No subscription.
30-day money-back guarantee
Your inbox has 47 unread tickets. You already know what 33 of them say. Same 20 questions, different customers. Your team is copy-pasting the same answers every day. That's the whole problem.
The AI Customer Support Agent reads incoming tickets, checks your knowledge base, and responds. For the hard 30%, it escalates to Slack with customer history, ticket context, and a suggested reply already written. Your team handles the exceptions. The agent handles the repetition.
What it does
Auto-response
Reads the ticket. Searches your knowledge base. Sends an accurate reply — without you touching it. Gets it right 85%+ of the time.
Smart triage
Sorts every ticket into 4 buckets: auto-resolve, needs context, escalate, or urgent. Zero manual sorting required.
Slack escalation
When confidence drops below your threshold, it sends the full file to Slack: customer plan, ticket history, and a suggested reply drafted for you.
Confidence scoring
Only fires when it's sure. Default threshold is 70%. Anything below that → escalated to a human. It never guesses.
Conversation logging
Every ticket logged — resolved or not. Weekly report: volume, resolution rate, and sentiment breakdown.
Here's what an escalation looks like in Slack
The agent never guesses. Low confidence = escalate with context, not auto-respond with wrong info.
Connects to
$99. Paid back in 2 days.
100 tickets/week × 5 minutes each = 8+ hours. At 70% auto-resolution, you get 6 hours back. Every week. Forever. That's $99 paid back before Friday of your first week.
Why not Intercom Fin?
| This Template | Intercom Fin | |
|---|---|---|
| Cost | $99 one-time | $0.99–$2.00 per resolution |
| At 500 tickets/month | $0/month | $495–$1,000/month |
| Annual cost | $99 total | $5,940–$12,000/year |
| Vendor lock-in | None — you own the code | Yes — their platform |
| LLM choice | Any (Claude, GPT, Gemini) | Fixed |
| Customisation | Full source code | Limited settings |
| Data | On your servers | On theirs |
| Email support | ✅ Built-in IMAP/SMTP | ❌ Intercom only |
What's included
28 files · 3,000+ lines · Zero npm dependencies · Pure Node.js
Running an agency?
Deploy for every client. Bill $3,000–$8,000 per setup. One purchase. Unlimited clients. No per-seat fees.See agency bundle →
FAQ
Does it work with just email (no Intercom)?
Yes. Gmail, Outlook, any provider with IMAP. Intercom is completely optional.
What LLM does it use?
Claude by default. Change one line in config.json to GPT-4, Gemini, or anything else.
How does the knowledge base work?
Drop .md files into the /kb folder. One file = one topic. Edit them anytime — no re-deploy needed. Or connect a Notion database instead.
What if it gets a question wrong?
It won't fire if it's not sure. Anything below 70% confidence gets escalated to a human. You control the threshold in config.json — lower it for more caution.
Does it handle non-English tickets?
Yes. Auto-detects the language, responds in kind. Tested with Spanish, French, and German.
What's the refund policy?
30 days. Email support@abbilabs.xyz. We fix it or refund it — no hoops.
Your team shouldn't spend 8 hours/week copy-pasting answers.
$99 once. Paid back before the end of week one. 30-day money-back guarantee.
Instant download · Full source code · 28 files · 3,000+ lines