Trigger

Trigger.dev is a managed platform for running long-running TypeScript tasks in production. It handles the infrastructure so you can write straightforward async code without worrying about timeouts, server provisioning, or queue management.

The core use case is AI workflows and agents. You define tasks that can call LLMs, use tools, process files, or chain multiple steps together. Tasks run with automatic retries, configurable backoff, and full execution logs. If something fails, you can replay individual runs from the dashboard.

Beyond AI, it covers scheduled jobs, media processing, browser automation, email sequences, and ETL pipelines. You can run Python scripts alongside your TypeScript code, add system packages, use FFmpeg for video tasks, or automate browsers with Puppeteer.

Realtime lets you stream task status and LLM responses directly to your frontend as tasks progress, without polling. If you want tracing and observability across LLM calls specifically, tools like OpenLIT or OpenLLMetry cover that layer.

Every run is logged with its inputs, outputs, attempt history, and timing. You can filter runs, apply bulk actions, and set up alerts via email, Slack, or webhooks when tasks fail. Each deploy is versioned atomically, so in-flight tasks aren't disrupted by new code.

Pricing is usage-based. You only pay for actual execution time. It's also Apache 2.0 licensed and self-hostable if you prefer to run it on your own infrastructure. For teams comparing workflow orchestration options, Temporal is a common alternative with a different approach to durability and state.