# How Exa Ships to Production Without Engineers Ever Pulling Code Locally

## About the company

Exa builds the search engine for AI, serving web search, crawling, and deep research through a developer API that powers Cognition, HubSpot, OpenRouter, and more than 400,000 developers.

Industry: AI Infrastructure**[Visit site](https://exa.ai/)**

Exa is building a search engine for AI, which means training models, crawling the internet, processing it, and serving it at scale.

Its engineering team works from the premise that every org is now a scaled org: a hundred engineers joined by hundreds of agents. These cloud agents onboard from scratch every single day in fresh sandboxes, and doing this is only worthwhile if the error rate of setting up a developer machine is driven to zero. That pushed Exa to make its repositories fully bootstrappable and automatable in the cloud rather than dependent on the accumulated local state of whoever set them up. Many of its newer engineers have never configured a local environment at all.

Devin works across the whole lifecycle, from refining ideas into specs, shipping deployments, and monitoring production metrics. In one instance, one multi-week Devin session fully built a test-system contract and rolled it out service by service across roughly 30 to 40 PRs.

> "If you want to parallelize more and you feel bottlenecked by the amount of tasks you can do in parallel end-to-end, then you need to go to the cloud, and you need to release some control, and Devin is the best platform to do that."
>
> João Adriano, Infrastructure Lead at Exa