Our vision
Every fleet should learn from its own experience.
We are building the infrastructure that turns a fleet’s real-world experience into its own intelligence. Our ambition is to make continuous learning a capability every Physical AI team can build on and own, not a system each must reinvent or rent from someone else.
Independent infrastructure for Physical AI
Emal AlwisFounder & CEOThe opportunity
An encounter should become an improvement.
Consider a vehicle approaching a construction zone that changed overnight. Its model struggles to interpret the new layout. The recording should become more than a log file or a one-off investigation. With the right context, it can help the team understand the failure, prepare better training data, and test whether the next model handles that situation more reliably.
That is the future we want for autonomous machines: experience that becomes useful learning, repeatedly. A delivery robot, a harvesting machine, and an industrial arm face different problems. Their builders share a need to connect what happens in the world to what changes in the model. We want that connection to become dependable infrastructure.
The founder's insight
The model is one part. The loop is the advantage.
At Tesla Autopilot, I worked on systems that turned fleet edge cases into model improvements. That work taught me to look beyond the model itself: how the team finds useful examples, preserves sensor context, decides what happened, and turns those decisions into the next training set. The ability to repeat that process matters as much as any individual breakthrough.
I founded Avala because other Physical AI teams should not have to build that entire foundation from scratch. They should be able to put their best engineering into the machines and models that distinguish them. Avala should give them a dependable way to learn from the world, from their first production workflow to the fleets they eventually operate.
The learning loop
From a real-world failure to a result you can verify.
The system we are building connects the full improvement cycle. Finding a failure is the beginning. Preparing the right data and measuring what changes are the work that follows.
Real-world experience
Find useful examples. Collect what is missing.
AI + expert judgment
Preserve sensor context. Resolve ambiguity.
Reproducible release
Pin the training data, labels, and quality evidence.
Train + evaluate
Test the next model against an agreed standard.
The foundation today
Start where the learning process breaks.
Our work starts with production data: multimodal sensor recordings, annotation, and quality review. MCAP and ROS recordings, camera images, LiDAR, and robot trajectories need to stay connected to the context that makes them useful. We bring software and expert operations together to prepare data that meets the customer’s training requirements.
Begin with a recurring problem: inconsistent labels, missing examples, or a training set that needs another round of repair. Define what acceptable means. Measure turnaround, quality, and total effort. Then make the next release easier to produce. That is how a specific data engagement can become a system the customer relies on, rather than another disconnected delivery.
Explore the platformWhere to start
Make the next training release count.
Avala is building a data improvement system that helps Physical AI teams find useful existing experience, collect what is missing, and connect each release to a result they can evaluate.
Start with a difficult capability, an underperforming dataset, or a recurring delivery problem. Establish the baseline. Improve the data. Measure the next model.
The gap
A capability map you define: where your current data already covers the work, and where it does not.
The release
The delivered recordings, annotations, provenance, and acceptance evidence that address the gaps you chose.
The result
A linked evaluation showing what changed, what did not, and what to acquire next.
We contract for the coverage and evidence we can verify. Whether the next release improves your model is established by an experiment you run, not promised in advance.
The decade ahead
Owners of their intelligence, not tenants of someone else’s.
By 2036, Avala is the open infrastructure on which any operator of machines in the physical world turns its own experience into its own intelligence, with data, models and deployment under its own control, so that everyone can be an owner of their intelligence, not a tenant of someone else’s. We intend to earn that position through four connected stages.
Prove one production workflow.
Solve a problem a customer needs solved now. Deliver against a clear acceptance standard and measure the result against the previous process. Make the next cycle more reliable, not just the first delivery impressive.
Become the foundation across models and teams.
Carry the same dataset identity, review history, and quality policies into more workflows. Each expansion should reuse the infrastructure already in place. The test is whether another team or model can benefit without another bespoke engineering project.
Automate the work between discovery and improvement.
Build toward agents that turn a training objective into an executable plan: find existing data, commission missing examples, coordinate expert review, prepare a release, and connect evaluation results to the next decision. Automate routine work while keeping consequential actions bounded, reviewable, and under customer control.
Make learning portable across the ecosystem.
Connect robot stacks, models, and infrastructure through open interfaces. Teams should be able to adopt better tools without abandoning their history. Shared datasets and models can extend that foundation where their owners choose to share and reuse is permitted. Interoperability should widen access, not require surrendering control.
The data that Physical AI will be built from does not exist yet. It is produced one deployment at a time, by the companies that own the machines, and it is mostly exceptions: the situations no one simulated. No single company will hold it. That is why the infrastructure that turns it into intelligence has to be independent, and why it has to leave ownership where the data came from.
What we believe
Progress should expand opportunity.
Customers should retain their proprietary advantage. Learning from a fleet’s experience must not mean quietly pooling its data with a competitor’s. We believe in portable data, clear rights, and customer-controlled deployment. Independence means your infrastructure partner is not owned by a model lab, a robot maker, or a hyperscaler. Avala should earn its place through the value of the system, not the difficulty of leaving it.
The people who teach and evaluate machines are part of this future. Their judgment deserves good tools, fair compensation, and paths to more skilled work. We want automation to remove repetitive effort while making expertise more valuable: better decisions captured in the process, better software supporting the next task, and fewer mistakes repeated.
The larger purpose is safer, more productive physical work and useful services that more people can afford. We want a small robotics team to build on capabilities that would otherwise demand a large internal organization. Better machines are the means. More people able to build, earn, and benefit from them is the goal.
Build with us
Build the infrastructure behind better machines.
We are starting with the data work that makes the next training release possible. We are building toward a world where learning from deployment is standard practice, not a capability a few companies can afford to build. Bring us a production workflow worth improving. Join us to build the infrastructure that makes the next one easier.
The longer-form essay behind this vision.