The fiftieth rejection did not sound new. Adam Cohen and Andrew Churchill had been pitching an early version of their company as “Gong for Engineers,” a system that connected development tools and found places where technical teams could improve. Chief technology officers liked the mockups. The meetings still ended on the same inconvenient question: how could the founders prove the product actually helped engineers?
They could not. Each lost deal made the answer more urgent. After roughly fifty versions of no, Cohen stopped hearing a sales objection and started hearing a product requirement. The industry did not merely lack coaching software. It lacked a credible way to describe how much engineering work was getting done. If the founders could solve that, the proof would no longer sit outside the product. It would be the product.
“The reason we lost every deal became our core product.”Adam Cohen
That reversal is the cleanest way into Cohen's story because it combines the parts of his career that might otherwise look separate. He is a salesperson who became an operator, an operator who became a software founder, and a founder who chose a measurement problem at the moment artificial intelligence made familiar measurements unreliable. Weave, the company he leads from San Francisco, now follows engineering work from prompt to production. It asks what the code accomplished, how much of it came from AI, what the work cost, and whether a cheaper model could have handled the task.
The wagon moves
Cohen grew up in North York, Toronto. His earliest business anecdote is almost suspiciously tidy: he put a lemonade stand on a wagon and took it door to door. He also remembers hustling friends to buy souvenirs on a school field trip. His father worked in venture capital and pushed him to succeed. Basketball and fantasy sports occupied the non-work side of his attention. Both reward a particular kind of mind, one that enjoys the contest but also wants to understand the roster, the matchup, and the numbers underneath the result.
At Western University, Cohen studied accounting and finance. Then came the less cinematic education of operating jobs. He co-founded BreVada and served as its chief operating officer. At Top Hat, he moved through national sales leadership and revenue operations before becoming head of operations and sales. He later led operations and finance at GlobalVision, then became vice president of operations and finance at Causal, the collaborative financial-modeling startup later acquired by Lucanet.
Causal also provided the consequential connection. Andrew Churchill was an early engineer there. Churchill brought the technical depth; Cohen brought years of watching organizations translate work into forecasts, quotas, budgets, and decisions. Together, they started WorkWeave in 2024. The name would shrink to Weave. The idea would change much more.
A product with too many sentences
The first WorkWeave launch promised personalized insights and feedback, a kind of always-on manager and career coach for engineers. The founders tried summaries. They connected Git and Jira. They built around the intuition that better feedback would improve performance. The product did not click. Three conversations could produce three different descriptions of what the company did.
The team applied to Y Combinator and got an interview, but not a place in the batch. Through the summer and fall, the founders kept pushing the feedback concept while sales stalled and fundraising went nowhere. Churchill later described waking up with the feeling that the business was failing. The hard part was not a lack of effort. It was that effort had no sharp edge.
Cohen and Churchill finally gave themselves a constraint: generate ideas they could explain in one sentence. “Quantify engineering productivity” survived the compression. This time, before their next YC interview, they launched and collected evidence that people wanted it. Weave entered the Winter 2025 batch. The acceptance mattered, but the more reusable lesson sits earlier. Clarity arrived after 134 engineer interviews, months of wrong turns, and enough lost deals to reveal a pattern. The short sentence was earned.
What counts when code is cheap?
Software teams have always struggled with measurement. Lines of code reward verbosity. Commit counts reward fragmentation. Story points can become a local currency whose exchange rate changes from team to team. AI makes each weakness more visible because a model can produce thousands of lines in seconds. A dashboard can show furious activity while the product moves sideways.
Weave's answer is to read the substance of the work. Its models examine pull requests, reviews, deployments, and the path from an initial prompt. The company normalizes contributions by asking how long the work might have taken in the pre-AI era. From that base, it reports output, AI return on investment, and opportunities for engineers to use agents more effectively. Its router chooses between models based on the job. A system-design problem may justify a frontier model. Centering a button probably does not.
By July 2026, Weave said it had analyzed more than two million human and AI contributions across more than 20,000 engineers at over 1,000 companies. Customers named by the company include PostHog, Telnyx, Reducto, and Robinhood. The last one became Weave's first Fortune 500 account and a useful test of Cohen's operating philosophy.
The Robinhood deployment required a self-hosted version and careful navigation through technical and organizational requirements. Churchill pulled all-nighters to build the deployment. A sales teammate stayed on daily calls. The entire Weave team joined a 15-minute standup for the duration, making the account everyone's problem. Cohen's notes afterward were practical: meet in person, find a champion inside the customer, and let the whole company feel the implementation.
Everyone sells
Cohen writes that “At Weave, everyone sells.” He does not mean engineers spend their days cold-calling. He means technical people should understand how the company captures value, while salespeople should understand the product well enough to protect what engineering can actually deliver. Engineers sometimes reach directly to technical leaders because they know the product. Sales handles the top-of-funnel disorder so builders can build.
The idea rejects the comfortable fiction that product quality and distribution live in separate departments. Cohen has spent too long in operating roles to believe it. A feature that never reaches a buyer does not become more valuable because it was difficult to ship. A contract that promises an imaginary feature is not a win. The organization works when the people making and selling the product can describe the same reality.
That ethic also explains Weave's current market. The company stands between engineering and finance, where an expensive new resource has arrived without mature controls. Coding agents consume tokens, developers choose among models, and executives ask what the bill purchased. Cohen expects a new operating function to emerge around the allocation of compute, tokens, and human attention. Weave wants to supply its accounting system and, increasingly, its routing layer.
“We started Weave to answer a question every engineering leader is now asking: what is our AI spend actually returning?”Adam Cohen
The next ten years
In July 2025, Weave announced a $4.2 million seed round led by Moonfire and Burst Capital, with Y Combinator participating. One year later it announced a $13.5 million Series A led by Standard Capital, joined by its earlier backers, IrregEx, and the Agent Fund. The new money is aimed at deeper models, a tighter loop between measurement and optimization, and support for larger enterprise deployments.
Funding provides time, not an answer. Cohen and Churchill have tried to make room for the longer question. At one point they disappeared into the woods for a day to think about Weave's next decade, away from customer calls, reviews, support tickets, and the weekly roadmap. The image is a useful counterweight to the company cadence: two founders who prize speed deliberately removing the immediate work so they can decide which work should exist.
Cohen's aspiration is broad. Weave should not stop at describing what happened. It should help teams decide which model receives a prompt, show engineers how to work better with agents, and connect technical output with business outcomes. That ambition will force the company to defend the quality of its own measurements. Any metric that enters management becomes an incentive, and any incentive can be gamed. The product must be useful to leaders without reducing developers to a scoreboard.
The childhood wagon still offers the cleaner metaphor. Cohen did not wait for customers to pass the lemonade stand. He moved the stand. At Weave, the motion came through a pile of rejection. Fifty failed sales calls carried the founders away from the product they wanted to pitch and toward the proof buyers needed. The noes did not become motivation. They became specification.