Fearn is a patent prosecution firm built for startups that need protection filed on the same timeline as their product releases. The company pairs former Big Law patent attorneys and PhDs with proprietary models purpose-built for patent law, and can take a founder from explaining an invention to a review-ready application in hours. Instead of hourly billing, it charges flat fees of $2,500 for provisionals and $9,000 for non-provisionals including USPTO fees , with provisional filings in as little as three business days and a guarantee that puts its drafting fee at risk if a non-provisional receives no allowed claims .
The firm runs on FearnOS, its in-house drafting and portfolio system. Rather than treating a patent as one long linear document, it represents the patent as a graph, mapping claims to the supporting text, figures and technical material behind them, while preserving attorney edits and a full record of how each section was produced. A human patent expert reviews every application, edits are protected from being overwritten by the model, and clients get controlled portfolio access, version history, and secure connections to where their technical documentation already lives.
Fearn was founded by Caltech alumni Han Kim and Angela Gao, who had seen the problem from opposite sides. Kim prosecuted patents at Morrison & Foerster, while Gao earned a PhD in computer science and AI. They initially built software for law firms but concluded that better tooling alone would not fix the model, since time saved meant revenue lost and sensitive pre-filing IP made AI difficult to deploy safely. The launch matters now because the firm is emerging from stealth with $5.5 million from Kindred Ventures, a16z Speedrun, Designer Fund and Essence VC and paying customers in software, hardware, defense and biotech, signaling a shift in how early-stage companies can buy patent work.
Comments (8)
Sign in to join the discussion.
P
Priya Balakrishnan17d ago
Flat fees on patent work is a real unlock, but I want to see the fine print on what counts as scope creep. Every firm swears they won't nickel and dime until claim 14 shows up.
T
Tomasz K.17d ago
The tweet cuts off right at 'Experienced patent' which is either a bug or the most cursed cliffhanger in launch history. Either way I clicked, so hats off.
M
Moses Adeyemi17d ago
ok wait, a firm building its own OS instead of duct-taping Clio and Word is the actual story here. FearnOS better ship as a product one day.
Y
yuki17d ago
how big is the team shipping both a law firm AND an internal OS? because that sounds like 4 people who haven't slept since March.
D
Demi Osabuohien17d ago
Every legal AI demo looks great until an examiner rejects the claims for indefiniteness. Curious what the review loop with the human attorneys actually looks like.
K
Kofi Mensah17d ago
Finally getting around to reading about 'AI patent drafting.' Is this the thing where the robot writes claims and a partner rubber stamps it, or has the pitch evolved?
R
Renata Villanueva17d ago
Flat fee model is bold when drafting time varies wildly by invention complexity. Either the margins are fat or someone is eating a lot of overage internally.
A
Arjun Mahadevan17d ago
Would love to chat about how FearnOS handles prior art search under the hood. DMs open if you're doing any press this week.
Work With Us
Book an intro call to explore our suite of services. We might have a waitlist.
Fearn
Big Law patents. Startup speeds.
Big Law patents. Startup speeds.
Tags
TLVC Rating
Community Rating
No ratings yet
to rate this launch.
Fearn is a patent prosecution firm built for startups that need protection filed on the same timeline as their product releases. The company pairs former Big Law patent attorneys and PhDs with proprietary models purpose-built for patent law, and can take a founder from explaining an invention to a review-ready application in hours. Instead of hourly billing, it charges flat fees of $2,500 for provisionals and $9,000 for non-provisionals including USPTO fees , with provisional filings in as little as three business days and a guarantee that puts its drafting fee at risk if a non-provisional receives no allowed claims . The firm runs on FearnOS, its in-house drafting and portfolio system. Rather than treating a patent as one long linear document, it represents the patent as a graph, mapping claims to the supporting text, figures and technical material behind them, while preserving attorney edits and a full record of how each section was produced. A human patent expert reviews every application, edits are protected from being overwritten by the model, and clients get controlled portfolio access, version history, and secure connections to where their technical documentation already lives. Fearn was founded by Caltech alumni Han Kim and Angela Gao, who had seen the problem from opposite sides. Kim prosecuted patents at Morrison & Foerster, while Gao earned a PhD in computer science and AI. They initially built software for law firms but concluded that better tooling alone would not fix the model, since time saved meant revenue lost and sensitive pre-filing IP made AI difficult to deploy safely. The launch matters now because the firm is emerging from stealth with $5.5 million from Kindred Ventures, a16z Speedrun, Designer Fund and Essence VC and paying customers in software, hardware, defense and biotech, signaling a shift in how early-stage companies can buy patent work.
Comments (8)
Flat fees on patent work is a real unlock, but I want to see the fine print on what counts as scope creep. Every firm swears they won't nickel and dime until claim 14 shows up.
The tweet cuts off right at 'Experienced patent' which is either a bug or the most cursed cliffhanger in launch history. Either way I clicked, so hats off.
ok wait, a firm building its own OS instead of duct-taping Clio and Word is the actual story here. FearnOS better ship as a product one day.
how big is the team shipping both a law firm AND an internal OS? because that sounds like 4 people who haven't slept since March.
Every legal AI demo looks great until an examiner rejects the claims for indefiniteness. Curious what the review loop with the human attorneys actually looks like.
Finally getting around to reading about 'AI patent drafting.' Is this the thing where the robot writes claims and a partner rubber stamps it, or has the pitch evolved?
Flat fee model is bold when drafting time varies wildly by invention complexity. Either the margins are fat or someone is eating a lot of overage internally.
Would love to chat about how FearnOS handles prior art search under the hood. DMs open if you're doing any press this week.