In today's tech landscape, a significant portion of code is now “AI-assisted.” Scott Guthrie, Microsoft’s Executive Vice President of Cloud and AI, noted in March that approximately 40% of the code developers upload to GitHub Copilot is “AI-generated and unmodified." This emerging trend has sparked new startups that aim to ensure AI-driven code remains effective and dependable — and investors are taking notice.
Recently, an Israeli startup called Digma secured $6 million in seed funding for a local continuous feedback platform that helps developers analyze their code, including AI-generated components, to spot potential issues. Similarly, Kolena, a San Francisco-based testing platform, announced a $15 million funding round to create tools for testing, benchmarking, and validating the performance of AI models.
Meanwhile, a nascent four-person startup in the Bay Area, Braintrust, introduced its own funding round of $3 million. Co-founder and CEO Ankur Goyal describes Braintrust as an “operating system for engineers building AI software,” designed to prevent poor outcomes from AI models. For instance, developers creating customer support chatbots can leverage Braintrust's technology to ensure accurate responses instead of generating misleading information.
Braintrust boasts a robust lineup of investors, including well-known angel investor Elad Gil, who played a vital role in incubating Braintrust's initial product. Other notable investors are Adam D’Angelo of Quora, Clem Delangue from the prominent AI firm Hugging Face, and Greg Brockman, co-founder of OpenAI. While the backing of such a prestigious investor group is promising, it remains to be seen if it will propel Braintrust to a leadership position in the competitive sector.
Goyal, who feels an innate connection to the mission of Braintrust, was raised in Pittsburgh by doctors and initially aspired to follow a similar path. His passion for technology emerged in high school during a linear algebra class where he learned about Google's PageRank algorithm—“I get goosebumps just talking about it,” he recalls. Transitioning from a medical career, Goyal pursued computer science at Carnegie Mellon University. He later dropped out during his junior year to develop a relational database system at MemSQL, a Y Combinator alum. After five years, he co-founded Impira and transitioned to head Figma's machine learning platform following its acquisition of the company.
Goyal’s experience gave him a deeper understanding of the complexities involved in building high-quality software in the AI era. Thus, he established Braintrust this past summer. “I spent considerable time developing traditional software,” he explains. “AI is different because it’s inherently non-deterministic. If you write AI code, you can’t guarantee its performance; real-world examples are essential for testing.” This process, known as evaluation or “evals,” presents challenges for companies lacking sufficient quality data for adequate testing. That's why Braintrust is collaborating with data-rich firms like Zapier and Coda, who are now beta testing their solutions.
“Their challenge,” Goyal highlights, “is that while they possess vast amounts of data and have many users, condensing that into representative testing examples can be daunting.” With Braintrust, he explains, “They can input all their data into our platform, run evaluations, and we help curate 'golden datasets' that assist them over time in assessing their software's effectiveness.”
As a key advantage, Goyal notes, “We integrate directly within their cloud environments,” streamlining compliance and promoting effortless adoption in enterprise settings.
While it's still early days and competition is heating up, with other startups like Deepchecks emerging with a focus on “continuous validation for AI,” Goyal underscores that Braintrust fulfills an essential need that he recognized at Figma—a solution that addresses gaps in the current AI software development ecosystem. “In the past decade, a science of software shipping evolved within continuous integration. However, until our launch, similar methodologies and workflows for AI simply didn’t exist.”
Pictured above: Coleen Baik (founding designer), Ankur Goyal (CEO), Manu Goyal (founding engineer), and David Song (product manager; part of Elad Gil’s team).
Ranking
