This Hacker Is Doing What Humane Couldn’t: Making the AI Pin Useful

This Hacker Is Doing What Humane Couldn’t: Making the AI Pin Useful

When the Humane AI pin was launched in November 2023, it promised to usher in a new future, where wearers could constantly interact with a powerful AI that could answer questions as the user went around the world.

By the time the $699 wearable chatbot began landing on buyers’ doorsteps in April 2024, the promise looked more prosaic. Bad reviews littered the web. Many, many people poked fun at it by comparing it—unfavorably—to the tech flop Juicero. To make matters worse, Humane warned users in June of that year that the pin’s charging case was a fire hazard.

Humane was purchased by HP last February, and most of the pin’s core features were shut down that same month. The device was doomed to become yet another contributor to the e-waste epidemic—except for some tech tinkerers’ attempts to give the gadget a second life.

One of those tinkerers is Adam Gastineau, a 29-year-old developer from Vancouver, Washington, who interned at Apple and has spent most of his career at Microsoft. He estimates he’s devoted around 400 hours so far to his project, which has resulted in PenumbraOS, an operating system that turns the pin into a development platform.

“The end goal is to make it so that other devs can use the pin to make whatever they want without worrying about any weirdness,” he says. “It just works.” To show its capabilities, Gastineau earlier this week posted to X a video of the retooled pin answering the question, “Who are you?” He also shared this clip with PCMag:



“The videos are the first demos of Penumbra doing something ‘new’ that’s useful to end users,” he says. “It shows the pin making a direct WebSocket connection to OpenAI’s real-time API, which takes in raw audio and outputs raw audio in close to real-time.” Read on to see how (and why) he did it.

Making a Teenage Dream Come True

Gastineau, who was obsessed with voice assistants like Apple’s Siri as a teen, says he’s “wanted a wearable that can gather context about your life for about 15 years now, but I knew I’d never really be able to make one.”

He says the Humane AI pin “looked cool,” but he didn’t like that it was locked down and couldn’t process AI requests directly on the device. But when he learned that Humane was shutting down its functionality, he bought one, hoping to add the features he wished were in the original gadget.

Gastineau benefited from an Android Debug Bridge (ADB) certification that leaked online shortly after Humane shut down its AI functionality. That ADB certification, which was controversial due to its provenance, “really demotivated Humane from helping us recover our bricked devices,” he says.

Get Our Best Stories!


Newsletter Icon


Your Daily Dose of Our Top Tech News

Sign up for our What’s New Now newsletter to receive the latest news, best new products, and expert advice from the editors of PCMag.

By clicking Sign Me Up, you confirm you are 16+ and agree to our Terms of Use and Privacy Policy.

Thanks for signing up!

Your subscription has been confirmed. Keep an eye on your inbox!


He sees a future in which people can load a private LLM onto the device, as well as a plug-in for a timer app that can utilize the pin’s built-in projector.

But it did help hackers like him who wanted to rejuvenate their pins. “Someone published a dump of most of the APKs”—Android Package Kits, which compile and contain all the files needed to run apps—“that same day, which can be decompiled into roughly the real Humane code, so I started pouring through them,” Gastineau says. Being able to reverse-engineer the code that makes the pin tick meant he was able to understand better how to code new functionalities for it.

Since then, Gastineau has been devoting “pretty constant work on the exploit.” He’s an iOS developer at heart and needed help in understanding how Android operates. So he used LLMs to translate his understanding of iOS into code that would work on Android’s internal systems. “I can do it the old way where I spend five hours scanning ancient StackOverflow posts for the right bit of info I need, or I roll the dice in 30 seconds on an LLM response,” he says.

Recommended by Our Editors

Creating a ‘Playground’ for AI Assistants

Despite all the hours he’s put into it, the project is still in its early stages. “All of this buildup is just for the ability to do ‘privileged’ things, like talking to the network,” he says. Gastineau can now activate alternative eSIMs on the pin, a big capability jump. The pin now is effectively “a normal Android device,” he says. “I usually like to ship extremely polished experiences, but I wanted to try to motivate people to start developing on the pin.” 

Gastineau’s long-term plan is to develop MABL, a central orchestrator app that will allow users to install different implementations of LLMs and other apps onto the pin. “The idea is to build a system that would let anyone replicate something Humane-like with their pin, in any way they imagine,” he says. “Basically make a playground for everything mobile AI assistants.”

He sees a future in which people can load a private LLM onto the device, as well as a plug-in for a timer app that can utilize the pin’s built-in projector.

cat image projected onto Humane AI pin user's hand

(Courtesy Adam Gastineau)

Look Out, Siri

Gastineau says he admires OpenPin, another project that enables Humane AI pin users to continue to use their devices post-mortem. But Gastineau wanted to do more. “OpenPin is basically the quickest and hackiest option possible,” he says. “It does literally nothing on-device except record and playback audio, take pictures, and detect touchpad touches. It doesn’t even use the laser display.

Developing an entire operating system for AI systems is tricky, expensive, and beyond the reach of an individual coder.  “Reaching the ‘intelligence’ level of Humane’s Cosmos [operating system] is not something we as a community are likely to do,” Gastineau admits. “[That] is frontier research and worth billions, probably.”  But he does think there’s a place—and much potential—to do something powerful nonetheless. “I’m here for a Siri replacement,” he says.

And he’s been happy with the community response. “The tweet has gotten a decent amount of attention,” he says. “I have had a few devs reach out to me, which is exactly what I wanted.” He’s also been heartened by the reaction from former Humane employees, who have seen and liked his posts.

“I want to try to drive the devs into a shared community so they can start talking,” he says. “Ultimately, I’m here to build things and make things work and have fun, so I don’t care that much,” he adds, “but obviously it’s cool to have people use your stuff.”

About Chris Stokel-Walker

Freelance Writer

Chris Stokel-Walker

Chris Stokel-Walker is a UK-based freelance journalist who focuses on technology. He is the author of five books and writes for The New York Times, The Washington Post, The Economist, Wired, Fast Company, New Scientist, The Guardian, The Times of London, The Telegraph, and many more.

Read Chris’s full bio

Read the latest from Chris Stokel-Walker

  • More from Chris Stokel-Walker

Leave a Comment

Your email address will not be published. Required fields are marked *