Fearn Raises $5.5 Million Seed to End the Two-Tier Patent System

Funding led by Kindred Ventures, with participation from a16z speedrun, Designer Fund, and Essence VC, will expand Fearn's AI-native patent platform built for a first-to-file world

SAN FRANCISCO, June 10, 2026 /PRNewswire/ -- Fearn, the AI-native patent platform, today announced a $5.5 million seed round led by Kindred Ventures, with participation from a16z speedrun, Designer Fund and Essence VC. The funding will expand Fearn's purpose-built patent models and product capabilities — including automated labeled figure generation — that let any inventor, from a solo researcher to an enterprise R&D team, draft a Big Law quality patent. By cutting patent drafting from 50+ hours to just 20 minutes and reducing costs by up to 96%, Fearn lets any inventor file faster than even the best-resourced IP organizations today.

Every patent system in the world is now first-to-file, and the priority filing date is what matters — not who invented first, but who filed first. This has produced a quiet two-tier system. Inventors with access to elite in-house IP teams can file in 48 hours, while everyone else waits weeks. Those weeks make the difference between saying yes to a speaking invitation and turning it down, between publishing and staying silent, between closing a deal and watching it slip. Those weeks are, very often, the difference between winning and losing.

"Patents are how inventors turn ideas into lasting value, but in a first-to-file system, the priority date is everything, and access to fast priority dates has been gated by cost, complexity, and institutional connections for far too long," said Han Kim, co-founder and CEO of Fearn. "I spent years in Big Law drafting patents the traditional way before building Fearn, and experienced firsthand what the gap looks like. Innovation moves at one speed and protection moves at another, and that mismatch is where inventors lose ground. We built Fearn so protection moves at the speed of the work itself — leveling the playing field for a solo researcher and an enterprise R&D team alike."

Fearn's co-founders bring a rare combination of patent law experience and original technical research. Han Kim previously drafted and prosecuted patent applications across software, life sciences, and mechanical arts as a scientific analyst at Morrison Foerster and conducted bio-inspired neural algorithms research as a PhD student at Caltech. Co-founder and CTO Angela Gao completed her PhD at Caltech under Professor Katie Bouman — where she and Han first met — with peer-reviewed work on physics-aligned generative models published at NeurIPS, CVPR, and TMLR. She also developed proprietary AI models at Google Research and now leads Fearn's technical work.

Fearn is not a wrapper on a general-purpose model. The company's architecture composes dozens of small models — some LLM-based, some built entirely from scratch — engineered to resist hallucination and to produce drafts with formally verified guarantees against the failure modes that matter most in patent prosecution, including written-description sufficiency under 35 U.S.C. § 112(a) and the figure-labeling requirements of 37 C.F.R. § 1.84(p)(5) and MPEP § 608.01(g). Several of Fearn's models are built on a deliberate inversion of standard technical practice: rather than training on large, noisy corpora, they are designed to extract maximum signal from small, hand-corrected, expert-labeled datasets — the regime that truly matches what patent prosecution requires.

"Our world is accelerating due to speedups in technology and science, due to human ingenuity and AI model advancements. Intellectual property is the record-of-truth for technology, the design, engineering, and science carta for the world's industries, and the law and economics code by which we operate as innovators and capital markets," said Steve Jang, Founder and Managing Partner at Kindred Ventures. "To modernize this critical cortex of our innovation economy, Fearn is developing custom IP-specific models for deeptech, biotech, and hardware companies to quicken and accelerate their drafting and enforceability, and also a multi-model agent harness and patent drafting agent that orchestrates at the speed of GPU compute these once-antiquated and slow workflows."

"The massive wave of attention on our ambient robotics thesis and inaugural product, the Lume, meant we needed to move incredibly fast to protect our technology," said Aaron Tan, CEO and Co-Founder of Syncere. "Traditional patent drafting takes months of man-hours — and lots of precious capital — that a pre-seed startup just can't spare. Fearn changed the game for us, turning our engineering explanations into high-quality drafts in minutes and a fraction of the usual cost, allowing us to lock down our IP overnight and ensure it was fully protected."

Fearn makes zero external API calls to outside AI models. The company self-hosts every model in its stack, eliminating the indemnification and public-disclosure risks that have made enterprise IP teams wary of generative AI. Enterprise customers can choose customer-hosted VPC, single-tenant VPC, or multi-tenant VPC deployment; all customer data is held under AES-256 encryption, with SOC 2, ISO 27001, ISO 27701, and GDPR compliance.

Fearn is already trusted by engineers and scientists at companies including Unity, Dandelion Energy, Serova Bio, Capsule, Syncere, Sans Strings, Taya, and Cainex.

To learn more or start drafting for a flat $2,000 per patent, visit https://fearn.ai/.

About Fearn
Fearn is the AI-native patent platform built for a first-to-file world. Founded in 2025 and headquartered in San Francisco, Fearn pairs purpose-built, hallucination-resistant models with automated labeled-figure generation to take inventions from disclosure to filing-ready draft in minutes, at a flat $2,000 per patent. Inventors take the finished draft to the counsel of their choice. Fearn is trusted by engineers and scientists at leading startups and public companies. Learn more at https://fearn.ai/.

SOURCE Fearn