I recently sat down with Adam and Eito, the founders of Weave, to talk about what they’re building and why we were excited to lead their Series A at Standard Capital.
At a high level, Weave builds AI to measure and understand software engineering. Every company is now spending more on AI coding tools, frontier models, tokens, agents, and developer workflows. But most teams still do not have a good way to answer the most basic question: what are we actually getting for all of this?
Weave starts by integrating with a company’s repositories and coding tools. From there, it builds models that understand the work being done and estimate how long that work would have taken a human engineer in the pre-AI era. Weave then connects that output to AI spend, giving companies a real view into how many human-era engineering hours they are getting for each dollar spent on tokens.
The analogy I used in the conversation is that Weave is a little like a Fitbit or Apple Watch for AI-assisted software engineering. First, it helps you understand what is happening. Then, once you have the data, you can optimize.
That optimization layer is what makes Weave especially interesting. Weave is building a router that can sit inside tools like Claude Code, Cursor, or Codex and choose the right model for the task. If you are doing complex systems design, you may want the strongest frontier model. If you are fixing a simple UI issue, you probably do not. Weave can make those choices automatically, using data from prompt all the way through production.
This is becoming urgent. Companies are realizing that blindly sending every task to the most expensive frontier model is not sustainable. They want the quality and reliability of the best models when they need them, but they also want the cost, security, and control advantages of open-source and self-hosted models where those are good enough. Weave gives teams a way to make that tradeoff intelligently.
The customer results are already compelling. Weave has seen companies cut AI costs by 20–80%, but the more important outcome is output. Teams using the router are seeing 20–25% increases in engineering output, and in some startup environments the speedups can be much larger.
The customer use cases also show how broad this category can become. Robinhood uses Weave to understand how AI is contributing to engineering work and where token spend is producing results. Telnyx, which has roughly 150 engineers and a very flat engineering organization, uses Weave to help engineers see their own data, improve how they work, and coordinate without relying on traditional layers of management. PostHog is also using the product.
Adam and Andrew came to this problem from a very specific insight. At their previous company, engineering was highly analytical in how the work was done, but surprisingly vibes-based in how the work was evaluated. Sales was almost the opposite: relationship-driven in the work itself, but extremely quantitative in measurement. Weave was born from the idea that engineering should get the same level of quantitative feedback and operational visibility.
Their first idea was closer to an AI engineering manager. That did not work. But when they tried to sell it, they kept hearing the same question: how would we know whether it is working? That led them to the deeper problem: companies did not have a good way to measure engineering output in the AI era. Counting lines of code, PRs, or story points was not enough. LLMs created a new possibility: actually understanding the work itself.
The timing also matters. Adam and Eito went through YC in Winter 2025, right as Cursor and AI coding tools were becoming central to how the best startups were building software. They were watching the future unfold in real time. The best teams were not just getting 10% better. Some were moving multiple times faster. But that only made measurement more important. If AI is changing engineering this much, every serious company will need to know what is working.
We also talked about team and culture. Weave is building a product for elite engineering teams, so the company itself has to operate like an elite engineering team. They are fully in-person in San Francisco, five days a week, and they use their own product to benchmark and improve how they build. The pitch to candidates is clear: if you want to work on hard problems at the frontier of software engineering and AI, this is a team where you can do that.
Weave is on the right side of a major shift in software engineering. AI is not just changing how code gets written. It is changing how engineering teams are measured, managed, and optimized. We are thrilled to be partnering with Adam, Eito, and the Weave team as they build the system of record for AI-era engineering.