French

We raised a $150M Series C at a $3B valuation, led by @sequoia, to give everyone in the enterprise the power to build and automate like an engineer. In under two years, AI coding tools have transformed how engineering teams work. Engineers describe what they want and ship it the same day. We want the same for everyone. From IT and HR to legal, security, talent, marketing, and beyond. We’ve onboarded two of the Fortune 20 in the last six months and doubled ARR every quarter since our Series B. Customers include @FOXCorp, @WPP, @LiveNation, and @tryramp. Now the hard part. Bringing @getserval to every employee means: → Trust at scale: agents never get more power than the person asking → Reliability for people who don’t write code → Agents that find and fix problems before anyone realizes something is wrong → One system of record for every department In the last 12 months, the team has grown by 10x. Now we’re hiring the best engineers in the world to build what’s next. If you want to work on these problems: serval.com/careers

When an agent can open dozens of pull requests a day, human review alone becomes a bottleneck. Security needs to keep pace with how we’re building. Sam Stewart (@nfFrenchie), our head of security, has been working on exactly this: if agents write the code, security requirements need to live where agents can read them and check their work as they go. So our threat models now live in the codebase. They evolve alongside relevant code changes and are checked in CI against shared security requirements. Sam also contributed a new Invariants feature to the open source tool threatcl, so agents can validate their threat models against those requirements before a person reviews the change. It’s the same principle behind @getserval: give agents clear boundaries and checks so they can take on more work safely. Proud of Sam and the team for this work. If you’re figuring out how security keeps pace with AI-assisted development, you can read his write-up here: serval.com/blog/agentic-thre…