The team is the credential
Built by the people who built the category.
Three problems have to be solved at once: live per-speaker attribution, regulated-vertical product judgment, and a patent strategy that arrives before the competition. The next names on this page are the point of this website.
- Employee #10 at Otter.ai, 2017–2025. Built Otter's speaker diarization and speaker recognition from zero, then led the streaming, real-time version. Diarization means telling speakers apart; it is the single hardest problem in our product, and he built the industry's reference version of it.
- Named inventor on 12 of the 19 Otter patent filings in the public record. The technical foundations of meeting AI have his name on them, verifiable at the patent office.
- Wake words and voice verification at Knowles/Audience before that. “Cull.ai, stop” must work on-device, in noise, instantly. He has already shipped exactly that class of system in embedded, real-time environments.
- LLM evaluation and hallucination-reduction work at production scale; UCLA M.S. The judgment engine is an LLM system whose mistakes have legal consequences, and he has built the machinery that catches them.
- Perhaps a few dozen engineers worldwide have shipped streaming diarization in production. One of them is on this cap table, building it clean-room from the first line.
- Stanford PhD in Electrical Engineering, M.S. in Financial Mathematics, Stanford Medicine postdoc. Research depth across the exact stack we need: signal processing, machine learning, artificial intelligence and the quantitative fluency our financial-services & healthcare buyers speak.
- Previously Co-founder and Chief Science Officer at an a16z Speedrun company (real-time audio AI). He built the evaluation frameworks and fine-tuning pipelines there, and evaluation is precisely what makes a consent gate trustworthy enough to sell.
- Shipped regulated AI three times: ZONARE Medical Systems, Vave Health, and RadSupport (as CTO). He designed HIPAA pipelines as a practitioner. Healthcare is our second vertical, and he has already built in it.
- 13 US patents filed, 400+ citations. He reads and shapes claim language with our counsel, which matters in a company whose moat is a patent family.
- His product demos put top hedge funds at $1,000 per user, per license. Financial services is our beachhead, and he has already demonstrated sales pitch for advanced technology at premium pricing.
- Inventor of Nextdoor.com and founding CEO of its predecessor company, which he took from inception through Seed, Series A, and Series B funding before its exit to Google. Nextdoor licenses his patents today.
- Founded Trademarkia.com and scaled it self-funded to $13M ARR, 125,000+ clients, and more than a million organic visitors a month, one of the largest IP platforms in the world, transformed from a 300-person labor model to 86 people running AI agents, with EBITDA improving as headcount fell.
- Five years as outside patent counsel to NVIDIA, authoring 100+ U.S. and international applications across AI, GPU architecture, and gaming.
- Patents he has written or invented have been licensed or sold for over $1 billion. First-chair federal litigator with 40+ cases; runs a 50-attorney law firm. California licensed attorney.
- The patent moat, the litigation readiness, and the studio capital behind Cull are not vendors we hired. They are his day job.
Why this specific team wins this specific market
Three problems have to be solved at once, and most teams can solve one of them. Live per-speaker attribution inside a buffer window: Hitesh built it, zero to one, at the company that defined the category, and it is not a skill you can hire on the open market. Regulated-vertical product judgment, from healthcare to finance: Matt carries research credibility and enterprise-buyer fluency in the same person, the rare combination this category requires. And a patent strategy that arrives before the competition: the studio filed six accelerated applications before the market noticed. The gaps interlock rather than overlap.
The studio and the bench. PatentVC, the venture studio of LegalForce RAPC Worldwide, funds Cull from operating profits and has already done the work most seed companies spend two years on: patents filed, name and mark cleared and owned, corporate structure papered, and capital committed as needed to reach the next stage. Behind the founders sits an engineering bench in Bangalore, built around the CTO in his time zone and hired against a bar of shipped real-time audio, with architecture review from senior Trademarkia engineers, a cost structure that buys two to three times the engineering of a Bay Area team. The studio's founder previously built Fatdoor/DealMap, sold to Google, with the concept restarted as Nextdoor, which licenses his patents as a public company today.