Coval is evaluating AI-based voice and chat agents such as self-driving cars
What do voice AI agents and self-driving cars have in common? Their performance can be evaluated in the same way, says Brock Hopkins, a former technology leader at Waymo. Coval, Hopkins’ new startup, is looking to do just that.
“When I left Waymo, I realized that a lot of these problems we were facing at Waymo were exactly what the rest of the AI industry was facing,” Hopkins (pictured above, center) told TechCrunch. “But everyone was saying that this is a new paradigm, that we have to come up with testing practices from first principles and that we all have to recreate everything. And I looked at that and said, wait, we’ve spent the last 10 years in autonomous driving to figure out how to do this.”
In 2024, it decided to launch it Covalenta platform that creates simulations of AI-powered voice and chat agents that test and evaluate how they perform tasks in the same way Hopkins tested Waymo’s self-driving cars. Coval can run thousands of simulations simultaneously, such as asking an agent to make a restaurant reservation or asking an agent to answer a customer service question asked indirectly.
Coval’s technology evaluates proxies based on a common set of metrics, but companies can also customize what they’re looking for and use Coval to further evaluate regressions. Users can also take this data and the insights that shine from it and present it to their end customers for either a demo or as a monitoring tool to show their customers that the agent is working as intended.
“One of the biggest barriers to companies adopting agents is that they feel confident that this isn’t just a smoke-and-mirrors demo,” Hopkins said. “Choosing between vendors is a really complex task for these executives because it’s very difficult to know what to ask for or how to even prove that these agents are doing what you expect. This gives our companies the ability to really show that and demonstrate that.”
Hopkins actually formulated the idea behind Coval during Y Combinator’s Summer 2024 push before launching the product publicly in October 2024. She said demand has been strong and has become explosive in the past couple of months, with clients wondering how quickly they can get their agents evaluated.
The San Francisco-based startup is now announcing a $3.3 million seed round led by MaC Venture Capital with participation from Y Combinator and General Catalyst. The startup will use the capital to build out its engineering team and work toward product-market fit. Hopkins added that the company will also enable its users to evaluate other types of AI agents, such as web-based agents, in the future.
Koval appears on the scene just as momentum — and hype — around AI agents appears to be at an all-time high. Technology leaders in organizations love Marc Benioff They praised (and marketed) the technology by saying Salesforce will deploy more than a billion AI agents by next year. OpenAI It is rumored that his film about an AI agent will be released very soon.
There are also many startups being created in this field as well. There were more than 100 startups working on building AI agents across Y Combinator’s three groups in 2024 alone. Some AI startups have received large venture funding rounds as well. one, /dev/agentsraised a $55 million seed round at a $500 million valuation in November 2024, less than a year after its founding.
This momentum means that there will likely be more companies looking for help evaluating their agents as well. Hopkins said Koval has a good chance of standing out from the pack because, unlike the inevitable newcomers, Koval has a head start.
“I think what really sets us apart is that I’ve been in this field for half a decade and have built these systems over and over again,” she said. “We’ve built multiple iterations and seen how they fail and how they scale, and we’re building the same concepts into Coval and all those learnings.”
