Trigger.dev Rebuilt Its Product to Kill Timeouts

The story

Trigger.dev was founded by four people who had already been working together for about seven years: Matt Aitken, Eric Allam, Dan Patel and James Ritchie. Y Combinator's profile of the company lists their earlier credentials. Allam had been CTO of Code School, an early platform for learning to code that was acquired in 2015. Aitken had created two iPad apps that won Apple's iPad App of the Year award. Together, the group had built JSON Hero, an open-source JSON viewer that YC said was used by more than 35,000 developers a month. The company was founded in 2022 and joined YC's Winter 2023 batch, with Aitken as chief executive and Allam as chief technology officer.

Aitken told the origin story himself in a February 2023 post on Reddit's web development forum. He wrote that he and three friends had built jsonhero.io during evenings and weekends, and that to their surprise it quickly became popular as a way to view JSON documents, collecting thousands of GitHub stars within weeks. That success made them want to follow it with a related developer tool, because JSON Hero was free and was never going to be a business on its own. After many ideas and some failed attempts, they settled on Trigger.dev. Aitken explained that the team liked Zapier but believed tools of that kind fell short for developers.

The sequence after that moved quickly. According to Aitken's post, they applied to Y Combinator and were accepted, quit their jobs, built a beta in a few months, and announced it on Hacker News, where it reached the front page and received more than 2,000 GitHub stars in 48 hours. The title of the post said YC had invested $500,000 in the project.

YC's launch announcement in February 2023 described the first version as a developer-first, open-source alternative to Zapier that let developers create workflows directly in code. The example was mundane on purpose: when a Stripe payment fails, create a Linear issue and email the payments team. The founders were quoted saying that tools such as Zapier, n8n and Make were fine for simple tasks but unsuitable for advanced use cases, and asking why developers should have to leave their codebase to write such workflows.

That first product, later called version 2, had an architectural constraint that turned out to define the company's next phase. In v2, a customer's workflow code executed inside the customer's own serverless functions. Serverless platforms impose time limits, so Trigger.dev made long-running work durable through a trick: it cached the completed chunks of a function and replayed the function repeatedly until all the work was finished. The company's v3 announcement acknowledged that this approach had key downsides that could not be fixed as long as code ran in customers' serverless environments. Developers had to wrap actions in special calls so results could be cached for replays, and they often relied on Trigger.dev-specific integrations rather than ordinary SDKs.

When v3 later opened to all users, the company summarized what it had learned from speaking with hundreds of v2 users. The first problem was timeouts: splitting jobs into chunks was possible but the code was hard to write and confusing. The second was very long processes, such as video transcoding and some AI operations, which could exceed even generous serverless limits like those of AWS Lambda. The third was double billing, because customers paid Trigger.dev for orchestration and their hosting provider for compute. The team concluded that eliminating timeouts required building its own infrastructure to host, manage and scale customers' tasks.

This was a significant bet for a small startup. Instead of acting as an orchestration layer over other people's compute, Trigger.dev would deploy and run customer code itself. The developer preview post explained how. Tasks run on long-running servers, and when code waits, schedules a retry or waits for subtasks, the platform can freeze it and resume later. The underlying technology is CRIU, short for Checkpoint/Restore In Userspace, which the company noted had been used at scale by Google since 2017. Each run has its own process that survives freezing and unfreezing, which Trigger.dev called stateful serverless. On the cloud product customers pay only while code is executing, not while it is paused.

The v3 announcement presented the result as a simplification for developers. They could write normal asynchronous code, use regular SDKs and HTTP requests, and stop thinking about timeouts because there were none. Tasks became the core primitive instead of jobs. Costs also became simpler, since Trigger.dev now provided both compute and orchestration. The company said it remained committed to open source and self-hosting.

The rebuild positioned the company for the wave of AI applications that needed exactly these properties: long-running chains of model calls, retries when external APIs fail, queues to manage concurrency and observability to understand what happened. YC's current profile describes Trigger.dev as an open-source platform for building AI agents and workflows in TypeScript, with long-running tasks, retries, queues, observability and elastic scaling.

In December 2025 Aitken announced a $16 million Series A led by Standard Capital, a new Series A firm led by Dalton Caldwell, Paul Buchheit and Bryan Berg; Trigger.dev said it was part of the first cycle of Standard's inaugural fund. Michael Grinich and CTO Fund invested for the first time, and existing investors including Y Combinator, Liquid 2, Wayfinder Ventures, Pioneer Fund and Rebel Fund participated again. The company said more than 30,000 developers were using the platform to execute hundreds of millions of agent runs each month, citing customers such as MagicSchool, Icon.com and DavidAI.

The roadmap in that announcement focused on more advanced observability, faster run starts using MicroVMs, more third-party integrations, sandboxes for executing untrusted code and tools for context engineering. Aitken also described the company as a US-style startup based in Europe and said it was hiring engineers. The Silicon Valley Post's report on the round repeated the same plans and listed the four co-founders.

Trigger.dev's history shows a founding team willing to discard the architecture that got it its first users. The replay model was clever and let the product launch quickly on top of other people's infrastructure. Listening to users made clear that clever was not enough, and the company took on the far harder job of operating compute itself.