Generative AI has become a key part of coding workflows, but most companies struggle to track its use, let alone its return on investment. Israeli startup Milestone hopes to help with a platform designed to correlate AI tool usage with engineering metrics, including code quality.
The catch is that these companies have to give Milestone access to their codebases, a bet that investors initially questioned, CEO and co-founder Liad Elidan told TechCrunch. But with customers, including Kayak, Monday, and Sapiens, the startup has now raised a $10 million seed funding round led by San Francisco-based venture firm Heavybit and Israeli fund Hanaco Ventures.
In an unusual arrangement, Elidan and Milestone’s CTO, Professor Stephen Barrett, had gone years without meeting in person by the time they started fundraising. Unlike most of Milestone’s Israel-based team members, Barrett lives in Ireland and teaches computer science at Trinity College Dublin, where Elidan was once his student, and the two bonded through software projects.
Despite the distance, the duo kept in touch over the years and eventually decided to found a startup focused on engineering efficiency, just as coding assistants and other code-generation tools were taking off. GitHub Copilot has since crossed the bar of 20 million users, but companies still lack visibility into how these tools are used and are impacting productivity.
According to Elidan, Milestone answers these questions by relying on four pillars — codebases, project management platforms, team structure, and the codegen tools themselves — to create what he describes as “a GenAI data lake.” In practice, this gives organizations actionable data on which teams use AI and to what effect — thanks to their own information.
Armed with this data, managers under constant pressure to leverage AI to push productivity are, for instance, able to measure feature delivery speed, Elidan said, and can find out whether recent bugs were caused by AI-generated code and can make informed decisions on where to implement these tools.
This also gives Milestone a front-row seat on ROI — the “holy grail question” that it aims to answer granularly for its customers. But on a high level, he said, “We don’t have a customer that used Milestone and said, ‘Okay, GenAI doesn’t help me, I’m going to revoke all my licenses.’ It’s actually the opposite. They want to try more Gen AI tools.”
This fast-paced adoption also means that Milestone has to keep up with a rapidly evolving landscape. “It used to be auto-completes, then it was chat, then it was agentic-based chat, and it keeps going,” Elidan said.
That’s also where Barrett’s academic background helps the team understand the wave of transformation its customers are going through. “A lot of the ways we used to think about engineering are going to have to change,” the professor told TechCrunch. “I think in some sense, AI is filling out the team, and engineers are now becoming managers.”
To keep up with the tools that power this wave, Milestone says it has partnered with many vendors, such as GitHub, Augment Code, Qodo, Continue, and Atlassian — the company that powers Jira and whose venture arm Atlassian Ventures also participated in this seed round.
The round was also supported by angel investors, including GitHub co-founder Tom Preston-Werner, former AT&T CEO John Donovan, Accenture’s senior tech advisor Paul Daugherty, and Datadog’s ex-president Amit Agrawal — who all understand that what Milestone is building is relevant for the enterprise market, Elidan said.
That enterprise focus was deliberate from day one, with Milestone even saying no to prospective clients that were too small — “a very hard thing to do,” Elidan said, but one that gave the startup clarity around a roadmap requiring enterprise credentials and features. Focus would be his main advice to other founders, and Milestone is taking it: The startup won’t expand into measuring GenAI’s impact on marketing or other functions, even as it grows.
