My next step: joining Calibrate Ventures to invest in Embodied Intelligence

6 minute read


After 7 wonderful years leading ML at TRI, it is time for me to embark on a new adventure: I’m thrilled to be joining the early-stage deep tech VC firm Calibrate Ventures as a Partner! Why? Because Calibrate is uniquely positioned to back the most promising startups in Embodied Intelligence.

Why now? The Cambrian Explosion of Embodied Intelligence applications

We are on the cusp of a Cambrian explosion of robots and real-world AI applications, something my friend Dr. Gill Pratt foresaw back in 2015 for robotics. This is fueled by compounding technical breakthroughs in Embodied Foundation Models (EFMs, including VLMs and VLAMs like RT-2), sample-efficient policy learning (e.g., Diffusion Policies), and data collection (e.g., Aloha, Mobile Aloha, UMI) integrated with prior knowledge in what I call Principle-centric AI.

Pushing these scientific boundaries is something I still enjoy as a professor at Stanford and as a technical advisor at TRI. But I am also convinced this technology is about to blossom into many amazing startups.

I have already experienced this type of phase-transition four times throughout my career since 2007: as an AI researcher, engineer, manager, and executive in computer vision, machine learning, autonomous driving, and AI infrastructure. I know the thrill of discovery and building, but also working with visionary startups like Weights & Biases, Scale AI, and Parallel Domain as they went from 0 to 1.

Overall, the trend is clear and accelerating. My experience in academic labs, increasingly more applied R&D at the largest car maker in the world, and tight partnerships with startup pioneers convinced me AI is ready to get its hands dirty across many previously inaccessible applications. We are entering a new bountiful era of “market-ready moonshots” as we say at Calibrate. 🚀

The Cambrian explosion of Embodied Intelligence

Why VC? Growing the many branches sprouting from the AI trunk

Acting on this belief, I spent a few months exploring my own startup ideas (starting from a list of 112 😬). I quickly realized that many of them (well, maybe not all 112 of them 😉) seemed feasible (at least technically), exciting, and with high potential impact, but in very different directions. My personal mind map started to look like a hybrid between a giant sequoia and an orange tree: a massive trunk (AI) sprouting an explosion of fruit-filled branches (deep tech applications). The value is in the branches, each representing a different company.

The sequoia-orange tree of Embodied Intelligence

This is an image that can be generalized across the whole embodied AI field. There is a heavy focus on the trunk itself currently, and some even believe the trunk is all you need. I don’t know if AGI (Artificial General Intelligence) will ever be real, but I am sure there is no such thing as EGI (Embodied General Intelligence). The set of newly enabled applications of AI in the real-world is huge, and each requires unique companies (the branches) sprouting from common AI foundations (the trunk). The diversity from this embodied intelligence big bang is as inevitable and positive as the biological big bang was more than 500 million years ago. In a fascinating repeat of history, vision (biological and artificial) is a huge part of the trunk in both big bangs!

So why deep tech VC then? Because it is the best way to grow as many of these branches as possible! The goal is to build a strong diverse portfolio of startups uniquely enabled by revolutionary technology. This imminent Cambrian explosion of Deep Tech applications of AI in the real world is exactly why early-stage VC at Calibrate was the most logical conclusion for me.

Why Calibrate? Deep Tech applied at scale

Calibrate Ventures is an early-mover in deep tech investing with a growing portfolio at the forefront of vertical AI applications, AI infrastructure, robotics, and automation. As a Partner joining Co-Founders and Managing Partners Kevin Dunlap and Jason Schoettler, my role is to discover, fund, connect, and advise ambitious technical founders solving huge real-world problems with the help of cutting-edge AI. You can learn more in Calibrate’s official announcement.

On a personal note, I have had the pleasure of working with Jason and Kevin for years through the EON conference and Calibrate’s portfolio companies. They are astute investors, tech optimists, trusted advisors, and wonderful human beings. We share a common vision and values that make me truly grateful for the opportunity to work with them and the rest of the team.

What’s next?

I am excited to bring my AI background to Calibrate Ventures and help the next generation of bold founders build amazing AI-enabled products that move the world! I’m looking to meet many creative and visionary founders, so if this sounds like you, please reach out! I am also eager to partner with other investors sharing our vision, and help early adopters harness AI to be at the forefront of this new industrial revolution.

In addition to this exciting new adventure, I am continuing in my role as Adjunct Professor at Stanford. I feel incredibly privileged to work with amazing colleagues and students that constantly blow my mind with their creativity, work ethic, and energy. I am also a technical advisor at TRI to help their groundbreaking long-term research efforts in AI.

These responsibilities are synergistic with my Partner role at Calibrate Ventures: Embodied Intelligence is still a small world of passionate individuals with a strong community of shared interests. The teams I built and collaborated with are incredible, and I am still rooting for their success. I am excited to continue along our collective journey, ascending the sequoia-orange tree of embodied intelligence and its explosion of branches.

What do you think?

Will we see a wide variety of deep tech applications of AI in the real world soon? If so, which ones excite you? If not, is it because it is premature and thus much more research is needed, or is it because we will have a single winner-take-all EGI (Embodied General Intelligence)?

Let me know what you think on Twitter and LinkedIn.