- “Every single time an email comes in, it's sent to Claude, Claude reads it and drafts a reply — API calls are totally free.” — Stefano
The problem
Help Scout-style AI reply add-ons cost ~$45-49 per agent per month — "it gets expensive very quickly." Stefano wanted AI-drafted support replies inside his existing ticketing system without paying for an AI add-on seat.
The AI solution
His ticketing software, Freshdesk, exposes free APIs: one endpoint for drafting/adding content to tickets and another for actually sending emails. He uses only the first — every time an email comes in, the ticket is sent to Claude via API, Claude reads it and drafts a reply, and the draft is posted back onto the ticket as a note. Nothing is ever auto-sent; a human reviews the note and sends.
He piloted this for a single course when he launched a challenge, feeding Claude the landing pages and all the FAQs for that course, and the drafts were "pretty good." The next step is broadening it beyond one course — including technical issues — by having AI analyze his full email history, which "takes time and quite a few tokens, but it's worth it."
Tools & costs
Results
API calls cost nothing versus ~$45-49/agent/month for built-in AI reply add-ons. Pilot on one course launch produced good drafts once Claude had the landing pages and FAQs as context.
How to copy it
- Check whether your ticketing tool has a free API with an endpoint for adding notes/drafts to tickets.
- Set up a flow where each incoming ticket's email is sent to Claude via API.
- Give Claude the relevant context — landing pages, FAQs, policies — for the products customers write about.
- Have Claude's reply posted back as a private note on the ticket, never auto-sent, so a human approves before sending.
- Expand coverage by letting AI analyze your historical support emails course by course.
Source
AI Hangout — July 2026 (recording) · watch this segment on YouTube (28:04) →