Building DeskCrew, the thing I keep coming back to is the trust line. Most "AI support" either over-promises (auto-sends wrong answers) or under-delivers (a glorified FAQ).
I landed on AI-drafts-the-reply, human-approves-before-send but I'm genuinely curious where other people draw the line.
Maker here 👋 I built DeskCrew solo because every helpdesk I looked at was either $100+/seat or bolted AI on as an afterthought — and none were built for what's coming: AI agents that need to use support tools, not just chat with them.
So it's two things in one desk:
For your customers — a widget + shared inbox where AI drafts replies from your own knowledge base and a human approves before anything sends. Site, email, Slack & Discord in one place.
For AI agents — an open "door" where autonomous agents pay per action (USDC, on-chain) to file or resolve tickets, still human-approved.
Priced per resolution, not per seat — so it doesn't punish you for growing.
The whole thing — site, dashboard, API, and the embeddable widget — ships from one Next.js codebase on
Vercel, with the AI running through the Vercel AI Gateway. That's the reason one person could build this much, this fast.
Would genuinely love feedback — if you run support today, what's the one thing your current tool makes harder than it should be?