HOME

Autona Machines · Demo & Progress · July 2026

The Loop, Working.

Not a roadmap. This is what exists today: each stage of our pipeline running end to end, producing a real fine-tuned policy from a handful of synthetic demonstrations.

VLA DEMO

Our Fine-Tuned Model, Running Inside the Robot Brain

Our own vision-language-action model, fine-tuned on our own synthetic data, running closed-loop at 25 Hz inside our robot brain harness alongside frontier LLMs. The VLA handles contact-rich motion. The LLM handles planning, verification, and recovery. Plain English instruction in, joint targets out.

25 Hz Control frequency
14-DOF State/action space
Days Seed to policy
Zero Teleop hours
Single-object retrieval among distractors “deposit the battery in the right-hand box” Instruction phrasing the policy never saw in training. Blocks and ball left exactly where they were. The policy identifies the target by language grounding alone, ignores the four distractor objects, and places cleanly.
Selective sorting with drop recovery “place only the red cubes in their matching container” Mid-episode, a cube is knocked loose. The policy re-grasps and finishes: both reds in the bin, both blues and the battery untouched. This is the recovery behaviour trained explicitly via our failure-injection pipeline.
What makes this different from other demos. No teleoperation was used at any stage. No motion capture. No human demonstration on hardware. The seed trajectories were synthesised with inverse kinematics, exploded in simulation on real arm geometry, and fine-tuned into the policy you see running. The entire pipeline from seed to this clip took days, not months. And the clip is an unedited single rollout, not a best-of.

◊ ◊ ◊

HOW IT WORKS

The VLA Is a Hand, Not a Brain

The rest of the industry is betting everything on one giant model. We are betting on the architecture around it. That is the difference, and it is why we can ship with a small, fast, cheap policy while others need a foundation-model-scale VLA to do the same job.

What most teams are building

The dominant approach treats the VLA as the entire robot. One end-to-end model receives pixels and language, and it must plan, reason about failure, decide when to stop, handle novel objects, recover from mistakes, and produce smooth joint targets. All of that in a single forward pass. The result is a model that needs to be enormous, needs billions of tokens of training data, needs months of compute, and still fails silently when it encounters something outside the distribution.

When that model fails, there is nothing else. The company waits for the next checkpoint.

What we are building instead

Our VLA does exactly one thing: fast, smooth, contact-rich motion on chores it has been trained on. It runs at 25 Hz and it is very good at that one job. But it does not plan. It does not reason. It does not decide when to give up. Those are the harness's job.

THE HARNESS (the brain)

A frontier LLM that plans the mission, predicts what should change, verifies what actually changed, recovers from mismatches, and reports failures honestly. It has five tools at its disposal:

VLA policylearned, 25 Hz
Grasp + motionanalytic, exact
Perceptionwhere is the mug
Verificationdid it actually work
Written codecomposes the rest

The VLA is one tool out of five. When it has been trained on a chore, the harness calls it because it is the fastest, smoothest option. When it has not been trained yet, the harness reaches for analytic primitives, perception, and code it writes on the spot. The chore still gets done. And those runs generate the trajectories that train the policy for next time.

Why this matters

The giant-VLA approachThe harness approach (ours)
Model must plan, reason, recover, and controlModel only controls. Planning and recovery are the LLM's job
Needs billions of tokens and months of computeFine-tune on our own sim data in hours, on one GPU
Silent failures: model does not know it failedPredict-observe-compare catches every failure at runtime
New chore requires a new training runNew chore often requires only a new tool or a few lines of code
Every frontier LLM release is irrelevant to the robotEvery frontier LLM release upgrades the robot for free
No path to a chore the model was never trained onThe harness can compose primitives and code for any chore from day one
Reliability requires a better modelReliability is produced by the architecture, not waited for from the model

Claude Code did not make the underlying model better. It wrapped a single API call in a loop that could read files, run commands, look at what happened and try again. That wrapper is what turned a chat model into something that ships software. We are doing that for robots.The harness insight

Two paths to every chore, not one. If the policy is not good enough yet, the engine composes primitives and code and the chore still gets done. Those written runs generate trajectories that go back to stage 01, training the policy that eventually replaces them. The learned path and the written path are not rivals. The written path is how we ship in month one. The learned path is how the machine gets smooth in year two. Both exist from day one.

What this buys us concretely

◊ ◊ ◊

PROGRESS

What We Have Built

Every item below is shipped, running, and producing output. This is not a feature list of intentions. It is a changelog of things that work, with evidence attached.

COMPLETE

Simulation environment on real robot geometry

Full MuJoCo simulation of our dual-arm system with production URDF, meshes, inertials, and controller gains. Gravity compensation runs per physics step. State and action spaces are byte-compatible with real hardware, so nothing trained here requires a rewrite to deploy.

14-DOF state/action · Real meshes · Real gains · Camera rendering at 640x360

COMPLETE

Automated dataset generation pipeline

From a small seed of IK-synthesised trajectories, the pipeline explodes a task into thousands of randomised episodes: varied object layouts, camera angles, lighting, colours, friction, bin positions, and instruction phrasings. A quarter of all episodes are deliberate failure-recovery pairs.

Domain randomisation · Held-out prompt phrasings · Recovery episodes at 25% · Non-target preservation scoring

COMPLETE

VLA fine-tuning on our own data

An open vision-language-action backbone, adapted to our embodiment and our chore library. Trained on rented GPU from our own synthetic datasets, with immutable checkpoints at fixed intervals and a training gate that refuses to begin unless the dataset audit passes.

Open VLA backbone · Per-embodiment fine-tune · Fail-closed data audit · Hashed checkpoint retention

COMPLETE

Sealed evaluation protocol

Every checkpoint runs against locked, held-out test splits: unseen scene seeds, unseen instruction phrasings, a stability hold after success, and an out-of-distribution set that stays sealed until validation chooses a winner. Clean runs, recovery runs, and non-target preservation are scored independently.

Deterministic seeds · Post-success stability hold · Separate clean/recovery/preservation scores · OOD split sealed

COMPLETE

Agent harness with frontier LLM

The operator agent takes a mission in plain language, plans the approach, calls tools (including the VLA), watches the cameras, verifies results against predictions, and replans on mismatch. The LLM reasons; the robot moves. The agent is never on the 25 Hz control path.

Frontier LLM navigation · Predict-act-observe-compare loop · Tool-based architecture · Action guard on every command

COMPLETE

Agentic CAD pipeline (Modbot)

A whole robot generated as code from a written specification. Changes are diffs, variants are parameters, and no sketch plane is touched by a human. The same agents that run the software loop run the mechanical loop.

Full humanoid from spec · Gripper, joint drive, variant generation · Agent-driven · Zero manual CAD

IN PROGRESS

Sim-to-real transfer on physical hardware

The final step: the same checkpoint that passes evaluation in simulation, loaded onto the physical arm with no modifications. Because the simulation uses real geometry, real gains, and byte-compatible data conventions, the transfer requires no rewrite. Currently commissioning hardware for first deployment trials.

Same conventions, no rewrite · No remote operators · Real failure data feeds back to stage 01

NEXT

Expanding the chore library

Once the first task family transfers cleanly, the loop turns again. Each new chore goes from a small sample set to a deployed, evaluated policy in days. The pipeline is the product; the chore list is what it outputs.

◊ ◊ ◊

PIPELINE

The Flow in Detail

Each stage of the loop, with the specific decisions and tools that make it fast.

01

Sample

What happens

We generate the smallest credible seed for a new task. For the demos above, we used inverse kinematics: build the scene in simulation, define waypoints, and let the solver produce smooth joint trajectories. No hardware, no teleop rig, no motion capture stage.

Why this is fast

A handful of IK solutions take minutes to generate. They do not need to be perfect. They need to be good enough to be worth multiplying, because stage 03 is what makes them large.

Output

Tens of joint-space trajectories in the exact format real hardware produces. Byte-compatible from day one.

02

Seed

What happens

Those trajectories are loaded into MuJoCo on the production URDF: real meshes, real inertials, real controller gains, gravity compensation per physics step. The result is a small, clean, controllable dataset that behaves like real hardware data.

Why this is fast

The sim environment is pre-built for our arm geometry. Loading a new task is parameterising an existing scaffold, not building from scratch.

Output

A validated seed dataset: state, action, image, and language, all at 14-DOF and 640x360 camera resolution.

03

Multiply

What happens

The seed becomes thousands of variations. Object positions, camera poses, lighting, colours, friction, bin locations, and instruction phrasings are all randomised. A quarter of episodes are deliberately injected failures with recovery trajectories attached, so the policy learns what to do after a dropped grasp.

Why this is fast

Randomisation is automated and parallelised. Once the seed validates, the expansion runs unattended overnight. No human labels anything.

Output

A large training corpus with in-distribution and OOD splits, held-out prompt phrasings, and explicit recovery data. Ready for fine-tuning.

04

Train

What happens

An open VLA backbone is fine-tuned on the expanded dataset. Rented compute, one run, immutable checkpoints at fixed intervals. A fail-closed gate audits the dataset before step zero: if provenance or format checks fail, training never begins.

Why this is fast

We are fine-tuning, not pretraining. The backbone already understands vision and language; we are teaching it our embodiment and our chores. A single GPU run, hours not weeks.

Output

A series of hashed, immutable checkpoints ready for evaluation.

05

Prove

What happens

Every checkpoint is scored on locked evaluation protocols: deterministic seeds the training never saw, held-out phrasings of the instruction, a post-success stability hold, and a sealed OOD split. Clean completion, recovery, and non-target preservation are scored separately.

Why this is fast

Evaluation is automated and deterministic. Nobody watches a video and calls it good. The protocol runs, the numbers come back, the best checkpoint is selected on validation alone.

Output

A single winning checkpoint with quantified performance across every metric, and a sealed test score that was never used to select it.

06

Harness

What happens

The winning checkpoint is loaded into the agent harness as one tool among five. The frontier LLM takes a mission, plans the attack, calls the VLA for contact-rich motion, uses analytic primitives where no training data exists, verifies every action against a prediction, and replans on mismatch.

Why this is fast

The harness is already built. Adding a new policy is registering a new tool. The reasoning, planning, and verification infrastructure does not change per chore.

Output

A deployable skill: a policy wrapped in verification, recovery, and plain-language reporting, ready for hardware.

07

Deploy

What happens

Because simulation used real geometry and real data conventions, the command that trained on simulated episodes runs unchanged on hardware. What breaks in the real world is the most valuable data we own, and it feeds directly back to stage 01.

Why this is fast

No sim-to-real rewrite. No domain adaptation stage. No gap to bridge. The data format is the bridge, and it was built that way from the start.

Output

A robot performing the chore in the real world, with every failure becoming tomorrow's training variation.

◊ ◊ ◊

BY THE NUMBERS

Concrete Metrics

Not aspirations. Measurements from the system as it runs today.

MetricValueWhy it matters
Loop time: seed to evaluated policyDaysNot weeks, not quarters. A new chore can be trained and proven in the time it takes most teams to collect demonstrations
Human teleop data requiredZeroIK-synthesised seeds, simulation-expanded datasets. No teleoperation rig, no motion capture, no labelling team
Control frequency25 HzSmooth, contact-rich manipulation. Fast enough for dynamic grasping, slow enough for safe guard intervention
State/action dimensionality14-DOFSix joints per arm plus grippers. Byte-compatible between sim and real
Recovery episodes in training25%The policy does not only know how to succeed. It knows what to do after a failure, because we trained it that way
Evaluation protocolSealed, deterministicUnseen seeds, unseen phrasings, stability hold, OOD split. No cherry-picking, no eyeballing
Remote human operatorsZeroThe machine detects its own failures and reports them. No one in another country is quietly driving it

◊ ◊ ◊

NEXT

See It Yourself

We will walk you through the full evaluation protocol, play back any checkpoint, show the dataset provenance, and demonstrate the harness reasoning in real time. Nothing is behind a curtain.