You Want to Partner? Write some code.

OpenPilot as an Open-Source Wedge into the Automotive Industry

George Hotz started a company called Comma.Ai – their goal is: “solving self-driving cars while delivering shippable intermediaries.”

George is known for his hot takes on AV technology – making a $10k bet with Elon Musk about his Tesla “Full Self Driving” timelines, publicly making fun of Zoox’s robotaxi prototypes

…and declaring that self-driving cars are a “scam” (watch the talk here for some more nuanced perspective on that claim…)

Talk is cheap, but when his company of <20 engineers has also built a best in class driver assistance system – outperforming companies 100x their size – you need to listen.

Source: Consumer Reports (there might be another post in the future about their rating methodology…)

While these past results are impressive, in 10 years Comma will have arguably contributed their greatest innovation by presenting automakers with an opportunity to join the open source community.

All of these pull requests (open source software-speak for “suggested changes”) came from a developer in china who doesn’t work for They were merged into a codebase that is deployed to thousands of cars.

In a blog post a few months ago they offer a vision for where this is going:

Many of the drivers in Linux are maintained by the companies that make the hardware. We imagine a similar situation here, where car companies maintain and improve their brand port (the connection to Comma.AI’s self-driving software)

Or to put it simply (as they have on their website): “If Tesla is the iOS of Self-Driving, then we are the Android”

There are lots of interesting potential offshoots of this type of combinatorial open-source innovation, where one company doesn’t have to own the whole stack.

Here’s one: Automated Parking

Imagine a Comma.Ai – powered valet service that can connect with a network of sensors monitoring a parking garage. Vehicles could move at low speed, operate empty and park in tighter spaces.

It’s been tried already, and we’ve been tracking some early projects in the DIMO knowledge graph:

Despite early efforts from Bosh, Ford, and Daimler it’s hard to imagine automated parking scaling without a system that works across multiple types of car, and that’s before you get to the integrations with multiple different types of parking management systems for residential, commercial, or public lots.

Just as phone or app developers explore the design space of mobile with Android, other innovative services will pop up around the core vehicle platforms on top of Comma AI.

Carsharing, goods delivery, vehicle charging, mapping data collection, safety & regulatory applications would all be possible – and customizable – with an open autonomy layer that is integrated with the vehicle hardware.

At [DIMO] we’ve built a platform that tracks existing (and potential) connections between companies. Our goal is to provide a tool that enables anyone to building new [projects] based on existing [devices] (like Comma 2) and [functions] (like OpenPilot).

Drop us a line and let us know what else could be built on Comma AI?

Want to chat with George to discuss some ideas? We’ll put up the first $250 towards their $1,000 asking price for any DIMO members.

If you’re interested in working on DIMO email us ( or hop in the DIMO discord.