Companies are spending heavily on AI coding tools. Many still cannot explain what that spending is producing.
Weave has raised $13.5 million in Series A funding, led by Standard Capital, Y Combinator, Moonfire, Burst Capital, IrregEx and Agent Fund also participated.
Founded by Adam Cohen and Andrew Churchill, Weave measures engineering work across both developers and AI agents. Its models analyze pull requests, reviews and deployments, then estimate how long the same work might have taken a human engineer before AI coding tools became common.
The problem is that traditional metrics such as lines of code, commits and pull-request counts can reward volume rather than useful progress. That becomes more misleading when an AI system can generate thousands of lines in seconds.
Weave connects engineering output with token spending. Its platform calculates the estimated return from AI tools and routes prompts toward the least expensive model capable of completing the task. Complex architecture work may require a frontier model, while a minor interface change may not.
The company says its platform has analyzed more than 2 million human and AI contributions from over 20,000 engineers at more than 1,000 companies. Customers using its prompt router have reportedly reduced costs by 20% to 80%, while broader deployments have produced output gains of 20% to 25%. These results are company-reported.
The funding will support deeper measurement models, improved prompt routing and enterprise deployments. Weave’s larger challenge is proving that engineering output can be measured fairly without turning productivity data into another form of employee surveillance.
