Trigger.dev
Build and deploy fully‑managed AI agents and workflows
5.0•16 reviews•630 followers
Build and deploy fully‑managed AI agents and workflows
5.0•16 reviews•630 followers
Trigger.dev is the platform for building AI workflows in TypeScript. Long-running tasks with retries, queues, observability, and elastic scaling.
This is the 3rd launch from Trigger.dev. View more
Chat Agent by Trigger.dev
Launching today
Chat agent is a way to build durable AI chat experiences that run on a
machine with no timeouts and keep streaming through refreshes and
crashes. The machine sleeps when nobody's typing and wakes where it
left off, without you managing any state.
Keep the AI SDK you already use: streamText on the server, useChat on
the client. chat.agent slots in underneath as a transport and the API
route between them goes away. Every turn is traced: prompts, tool
calls, latency and cost.
Launch Team / Built With
![Trigger.dev Realtime AI [LW24]](https://ph-files.imgix.net/415afe48-f35c-4e64-8902-4b5994ca10c7.jpeg?auto=compress,format&codec=mozjpeg&cs=strip&fit=crop&frame=1&h=48&w=48)
Kilo Code
announced earlier this week, the @Trigger.dev team is riding the tailwinds today on @Product Hunt. S/O for this new launch, OSS ftw!
Trigger.dev
Hi everyone! I'm James, one of the founders of Trigger.dev.
Quick context if you haven't come across us: Trigger.dev is a platform for developers to build and run AI agents and workflows in TypeScript. Open source, Apache 2.0, no timeouts, and you can see every run in the dashboard.
Today I'm excited to launch chat.agent, a backend for AI chat apps.
Why we built it
Chat agents are usually built on a request/response endpoint, which is a bad fit. You get a timeout you have to work around and no memory between turns. So you write everything to Postgres, add Redis so the stream survives a refresh, and push the slow work onto a queue that you then have to coordinate. That's a lot of plumbing before your agent does anything interesting.
chat.agent gives every conversation its own machine instead. It lives for the whole conversation, sleeps when nobody's typing, and wakes up where it left off.
What that gets you:
No timeouts. In our production data 1 in 20 turns runs longer than 36 minutes, and that's fine.
Refresh mid-response and the stream picks up where your browser stopped reading. Close the tab, come back days later, it's still there.
Memory between turns is just variables. The sub-agent you spawned four questions ago is still there tomorrow.
Waiting is free. Pause for a human approval overnight and you're not billed for the wait.
Tracing and per-turn cost and token metrics, with nothing to set up.
Works with the AI SDK you already use. streamText on the server, useChat on the client, and the API route between them goes away.
It's been running in production since June, including at Arena.ai, and has handled millions of sessions and 84 years of compute so far.
I'd love to hear how you're handling this today, and if you've built durable chat by hand? I'll be in the comments all day.
Thanks!
Full technical writeup if you want the details:
https://trigger.dev/changelog/chat-agent
And the repo, Apache 2.0 and self-hostable:
https://github.com/triggerdotdev/trigger.dev