Analysis
Guava has raised $12.5 million in seed financing led by Acrew Capital, with participation from Flourish Ventures, WillowTree Ventures, Clocktower Ventures and existing investor Commerce Ventures, Pulse 2.0 reported. The Los Angeles-based company builds a developer platform for AI voice agents aimed at enterprise customer interactions.
What Guava does
Guava develops its own speech recognition, language model and speech synthesis stack as a single integrated system, rather than assembling a voice agent on top of third-party APIs. The approach traces back to Gridspace, whose technology -- trained on more than 10 billion minutes of live call-center conversations, according to Pulse 2.0 -- underpins Guava's models. The company's first public model, Daytona, launched in September 2026 and is already used by more than a dozen enterprise customers, according to Pulse 2.0. A software development kit supporting Python, TypeScript and Claude Code lets developers, per the company, build a working voice agent in under five minutes on Guava's free tier.
Guava funding history
This seed round is the first disclosed outside funding for Guava; current reporting does not state a founding date, a valuation for the round, or any prior financing beyond describing Commerce Ventures as an existing investor. Raised is not valuation here -- no source names a post-money figure for this round.
“The company's first public model, Daytona, launched in September 2026 and is already used by more than a dozen enterprise customers, according to Pulse 2.0.”
Who invested in Guava
- Acrew Capital -- lead investor on the seed round. Founding partner Lauren Kolodny said of the founders, "His co-founders have been building voice AI for over a decade, long before it was fashionable," per Pulse 2.0.
- Flourish Ventures -- participant; co-founder and managing partner Emmalyn Shaw said, "For voice AI to scale, businesses need code-level control and consistency."
- WillowTree Ventures, Clocktower Ventures -- participants named by Pulse 2.0.
- Commerce Ventures -- existing investor returning in this round; its earlier investment was not dated or sized in current reporting.
Chief executive and co-founder Omri Dahan, previously chief revenue officer at Marqeta, said, "We built Guava to give enterprises agents they can trust with high-volume, high-stakes conversations." Co-founder and co-head of engineering Anthony Scodary, who built the underlying technology at Gridspace with fellow co-founder Nico Benitez, said, "We train and serve every model ourselves, so we can tune the whole pipeline around one goal." Mike Piechna is also a co-founder.
Voice AI market and competitors
The global voice AI agents market was valued at $2.4 billion in 2024 and is projected to reach $47.5 billion by 2034, a 34.8% compound annual growth rate, per Market.us data cited by Ringly. Guava enters that market well behind better-funded rivals:
- Giga -- raised a $61 million Series A led by Redpoint Ventures, with Y Combinator and Nexus Venture Partners participating, according to Fortune.
- Vapi -- raised a $50 million Series B led by Peak XV in May 2026, bringing its total funding to $72 million after surpassing 1 billion calls handled, per Tech Economy.
Against that backdrop, Guava's $12.5 million seed is a fraction of what competitors have already banked, leaving the company to compete on technical differentiation -- owning its full speech stack -- rather than capital. What that comparison misses is that neither Giga's nor Vapi's post-money valuation has been disclosed, so it is not yet possible to say whether Guava's early backers got a better or worse entry price than investors in the larger rounds. Our funding tracker shows voice AI as one of the more crowded sub-categories in this year's enterprise AI fundraising wave.
What to watch next
The figure to watch is Daytona's customer count a year from now, not today's dozen -- Guava's vertically integrated model is more expensive to run than a wrapper product, and it will need rapid enterprise adoption to justify the cost of maintaining its own speech stack.