Guava, a developer platform for building voice agents, announced Wednesday a $12.5 million seed financing led by Acrew Capital. Flourish Ventures, WillowTree Ventures, Clocktower Ventures, and Guava’s pre-seed investor, Commerce Ventures, also participated in the funding round.
Guava’s first publicly available model, Daytona, launched in September. It is reportedly already supporting more than a dozen enterprise customers and powering tens of thousands of calls daily.
According to the company’s announcement on Business Wire, Guava builds on more than a decade of foundation model development by Gridspace, a voice AI research lab that trained its models on more than 10 billion minutes of live conversations in regulated call centers.
READ: Indian American startup raises $150K pre-seed to grow Telugu storytelling platform (May 21, 2026)
“It’s not every day you discover a frontier lab with a full team of top ML engineers in a market as big as voice,” said Omri Dahan, co-founder and CEO of Guava. “The team is extraordinary. Nearly every company we spoke with was dissatisfied with the voice AI they’d tried. It sounds fine in a demo but breaks down on real, high-stakes calls. We built Guava to give enterprises agents they can trust with high-volume, high-stakes conversations. Now we’re building the go-to-market team to take Guava to every enterprise and BPO ready to move voice AI from pilot to production.”
Guava was co-founded by Dahan, Anthony Scodary, Nico Benitez, and Mike Piechna.
Lauren Kolodny, founding partner at Acrew Capital, said, “We’ve known Omri for years, so when he decided to step back into an operating role, we paid close attention. His co-founders have been building voice AI for over a decade, long before it was fashionable, and with a truly opinionated approach. This incredibly strong technical foundation, coupled with Omri’s ability to find market signal and scale real businesses, was a compelling opportunity.”
While voice agents typically rely on different components supplied by multiple vendors and are largely controlled through prompts, Guava takes a different approach.
READ: OpenAI raises $11 billion at $1.25 trillion valuation, falls short of $30 billion target (November 7, 2026)
On the model side, the company develops and operates its own speech recognition, language models, and speech synthesis systems as a unified stack. This allows it to control the voice experience from the first word a caller speaks to the last. On the development side, Guava’s software development kit (SDK) is designed to help developers integrate voice capabilities directly into their applications using business rules and code, rather than requiring them to become experts in managing voice agents and prompts.
“When you stitch together other people’s models, you inherit their latency, their failure modes and their release schedules,” said Anthony Scodary, co-founder and co-head of engineering at Guava.
“We train and serve every model ourselves, so we can tune the whole pipeline around one goal: a conversation that feels natural and finishes the job. This funding lets us advance our state-of-the-art models into edge device applications, on-prem installations, and full-duplex interaction, where the agent can listen and speak at the same time the way people do.”