How Syncere built a real patent portfolio at pre-seed.

Aaron Tan was finishing a robotics postdoc at Stanford when he and his cofounder, Angus Fung, started Syncere. The company the duo is building centers around a bold thesis called "ambient robotics". In this worldview, AI and robotics are embedded into the everyday objects people are already familiar with—as opposed to "humanoid robotics", which supposes a humanoid embodiment for achieving useful tasks. Syncere's ambient robotics thesis is manifested in the Lume: a floor lamp that sits unremarkably in your living room, and when nobody's watching, folds the laundry. The Lume launched with critical acclaim, at 1.3 million views, and a flood of preorders. But virality cuts both ways: every new eye on the company was also that of a potential competitor. The inventions inside Lume sprawled across computer vision, control theory, AI, and inverse kinematics, and Tan knew from experience that any one of them was the kind of thing a rival could steal in a blink, if it wasn't protected on paper.

The trouble was the paper.

A top Silicon Valley law firm had quoted Syncere $10,000 to draft a single patent application. That number, in the world of intellectual property, is not unusual—it's actually a sweetheart deal. It's also, for a company at Syncere's stage, the price of a real strategic problem: the future is unknowable, and a startup with a fixed IP budget has to guess which of its inventions will matter most, protect that one, and leave the others exposed. The ideal answer is to patent all of them. At $10,000 a draft, no one at Syncere's stage can.

Then there was a second problem, one Syncere knew intimately from years of building hardware. The standard hardware-company patent workflow begins with an engineer disclosing an invention to a patent attorney—sometimes to both in-house and outside counsel. Several weeks later, a draft comes back, rewritten into legalese so dense it's nearly impossible to tell whether the attorney has grasped the technical substance or has merely produced something that looks like a patent. The drafts are always plausible. Whether they're correct is another matter. 

When Tan told Alex Levenko, Syncere’s founding hardware engineer, about Fearn, Levenko didn't believe it. A new tool, even a well-intentioned one, wasn't going to do better than the trained attorneys he'd worked with at top tech companies like Nuro, Faraday Future, and Siemens. And he had tried the legal tools built on top of Claude or ChatGPT. They all had a tell: confident-looking drafts that, when you actually read them, were full of issues his patent attorneys would have quickly caught. But Fearn was a different kind of thing. It wasn't a wrapper. Fearn had built their own models and hand-labeled data and was taking a different approach entirely. Instead of relying exclusively on LLMs, Fearn was using a compositional approach based on neurosymbolic architectures. And that was the first technical description Levenko had heard from a legal AI company that sounded like it might actually work. He decided it was worth a try.

He'd come in with low expectations. But right from the intake, Levenko was pleasantly surprised.

With Fearn, an inventor explains the invention in their own words, just how they would to a colleague. Fearn then reflects back in real time what it has understood. Levenko, watching this happen, recognized something he'd been missing for years. The process he dealt with at his past companies had always left him guessing whether the attorney on the other end of the disclosure had actually grasped the technical nuance. Here, the feedback loop was tight enough to answer that question while he was still in the room.

Excited, Levenko then tried feeding some casual photos of their hardware to Fearn. He immediately got back clean labeled lined drawings typical of a non-provisional patent. Levenko was even able to tweak the drawing labels to be exactly the way he wanted them to be. He could adjust the labels' positions, and he could change the language of each label, so that it was using his preferred technical terminology.

The draft came back in 22 minutes. Using Fearn's chat feature, Levenko started interrogating it. He was an engineer, not a lawyer, and he wanted to know whether the legalese in each claim actually tracked the technical substance of his invention. For every one of his questions, Fearn explained why the patent was drafted the way it was, and highlighted the specific claims and passages it was pointing to. Levenko was reading his own patent and following it.

"For what a single patent used to cost us, we can now build a real portfolio. That changes what's worth protecting."

— Aaron Tan, CEO, Syncere

Syncere has since drafted several patents with Fearn. Each was sent to their outside IP counsel for review. Each came back with little to no substantive rewriting—the kind of light pass that attorneys do on their own work, not the rebuild they'd do on a draft they don't trust. Each was filed. The portfolio is now expanding at a pace that, until very recently, was not available to a hardware company at Syncere's size. The traditional $10,000 ticket price, which had forced founders for a generation to pick one invention and live with the exposure on the rest, wasn't a limitation anymore.

"A top Silicon Valley firm quoted us $10K for a single patent draft. With Fearn at $2K each, we've filed three for the cost of not even one, and our outside IP attorneys reviewed and filed every one of them. I'd recommend Fearn to any technical founder at an early-stage robotics company who's been quoted $10K-plus for a patent draft and is wondering whether the price reflects the work or the firm letterhead."

— Alex Levenko, Founding Hardware Engineer, Syncere

If you're a hardware founder sitting on inventions you can't afford to protect, try Fearn at fearn.ai.