Company

The mind for every robot in the building.

Most robotics capital is pointed at the body or the model. Cereal starts somewhere else: the building, the people on its floors, and the record every robot has to keep to be allowed there. The same mind ports to every venue after that.

06 / Market positioning

Two very different bets on embodied AI.

Body-first

Billions into hardware and one form factor. Real markets, but the intelligence comes last and stays welded to one body.

Cereal, mind-first

One portable mind for the whole facility: identity, permissions, rounds, staging, safety, an attested record.

We're building the mind for every robot in the building.

21 / Founder-authored IP

The founders already invented the hard part.

Named inventors on 11 granted US patents between them, 8 of them the simulation-driven robotics family from Google X, Everyday Robots, and Google DeepMind listed below: the exact sim-to-real approach Cereal runs on. Every filing below links to the public record.

01 · The live loop

Simulation inside every control cycle.

The simulator is not a training ground left behind at deployment; it participates in every action the robot takes. The family was continued and granted under Google DeepMind in 2025: active, and still invested in.

US11938638B2·US12472630B2

02 · Fleet-scale simulation

Many robots, many permutations, one controller.

The virtual-environment infrastructure that made training and testing at Everyday Robots scale.

US12202140B2·US11654550B1

03 · Learning from operation

Imitation learning that refines control policies.

Training signal mined from real operation: the mechanism behind supervision that compounds into autonomy.

US12226920B2·US12377536B1

04 · Sim-to-real rigor

Keeping simulation honest about reality.

Re-simulating recorded episodes and injecting noise for robustness: the ancestor of Cereal's replayable, attested trail.

US12168296B1·US12214507B1

Inventor credentials reflect work at prior employers; the patents are assigned to those employers. Full record on the founder profiles: Matt Bennice · Paul Bechard.

20 / Founding bench

Interoperability × robotics × simulation.

Unusually complete for a robotics company.

Chief Roboticist · Robotics platform

ex-Google X / Everyday Robots.

Profile →LinkedIn ↗
CTO · Robotics & simulation

Google X · Wing · Everyday Robots · DeepMind · Stack AV · 1X · Cruise.

Profile →LinkedIn ↗
CEO · Interoperability & AI products

Analog · Firestarter · Pocodot/Cole.

Profile →LinkedIn ↗
Martin Watson
Advisor · Healthcare GTM

Former CEO, SeeChange Health · SVP, UnitedHealth · Chief of Staff, Aetna · led healthcare strategy, Fidelity Health Care Group · co-founder, HCG Secure.

LinkedIn ↗
Kjerstin Easton
Advisor · Perception & robotics

PhD Robotics, Caltech · Google X (perception lead, Everyday Robots; Tidal) · Apple · Applied Minds · Maxar · VP Software Engineering, Nauticus Robotics.

LinkedIn ↗
19 / The open platform

Every robot is a sensor on the real world.

Cereal opens its stack so others can build on it, train it, and, with each owner's consent, contribute scoped real-world data back. The fleet improves the fleet.

Like rovers mapping Mars, a fleet of working robots gathers grounded real-world data no lab can simulate, then brings it back to the site mind to make every robot smarter.

Build the mind for embodied AI.