From data to deployed robot.
Config builds the infrastructure behind Physical AI. Loop is one platform that turns real-world demonstrations into deployable, task-specific robot policies — six modules and two engines, used together or on their own.
The Platform
Six modules, one loop.
One platform, demonstration to deployment. Loop turns real-world demonstrations into task-specific robot policies through six modules. Activate only what you need — nothing forces a linear pipeline.
plan
Define the task. Prove it's feasible.
Scope the task and prove it can be done before committing downstream — sample delivery, a reference video, and a teleop test. An approval gate unlocks the rest of the pipeline.
collect
Capture demonstrations at scale.
Multi-camera UVC recording with voice-triggered capture and pre-flight health checks. Capture on the desktop Recorder or Loop Mobile, with Config's Data Studio operators or your own on-site team.
curate
Human QA, episode by episode.
Validate and annotate every collected episode so only clean demonstrations move downstream. Bad takes are caught before they ever reach a training run.
transform
Human data, robot-ready datasets.
Convert human demonstrations into robot trajectories for your embodiment, then build a training-ready dataset in one pass. Teleop skips conversion. Powered by the Conversion Engine.
train
Fine-tune CFG-1 to your task.
Adapt CFG-1, our foundation model, to your specific task and embodiment. Just 10 hours of teleop data typically reaches 80% success. Powered by the Policy Engine.
deploy
Run the policy. Operate the fleet.
Run your policy on the robot and operate the fleet — live inference and robot status, on-prem device management, network health, and rollout evaluation that feeds the next cycle. Inference under 50ms on a single RTX 5090, powered by the Policy Engine.
plan → collect → curate → transform → train → deploy · compose freely
The modules are the surface you work in; two engines do the heavy lifting on the GPU. The Loop client is the same whether an engine runs on your bench or on a remote server.
Policy Engine
Powers train + deploy · hosts CFG-1
The GPU server behind training and live inference. Hosts CFG-1, our foundation model, and serves it on your robot — running locally on an RTX 5090-class GPU or on a remote server, reached by the Loop client over an API.
Conversion Engine
Powers transform
The GPU server that turns human demonstrations into robot-ready trajectories. Owns the canonical list of supported grippers and embodiments, so converted data lands in the shape your robot expects.
End Effector
2-finger grippers
5-finger gloves
Target Robot
Operation
Data Type
End-effector form factor is your choice. Robot arms are interchangeable, and your gripper doesn’t have to match ours. Pick what fits your task, or talk to us about supporting your hardware.
Inside Loop
Collect to curate.
Capture multi-camera demonstrations, then validate and annotate every episode before it moves downstream.
Inside Loop
Train to deploy.
Fine-tune CFG-1 on your dataset, run it on your robot, and keep it improving — every correction becomes signal for the next cycle.
How Loop Works
One platform. One pipeline.
Six modules take you from a real-world demonstration to a deployed robot policy. Two engines do the GPU-heavy work. Use the whole pipeline, or activate only the modules you need.
01
plan
Define & prove the task
02
collect
Capture demonstrations
03
curate
Human QA
04
transform
Human → robot data
05
train
Fine-tune CFG-1
06
deploy
Run & operate
Loop is one platform — the client is the same everywhere, and any module runs on its own with data you already have. Catalog is a separate product: ready-made, pre-converted datasets you can buy outright without running any of the pipeline.
How Customers Engage
Find your starting point.
One platform, six composable modules — plus ready-made datasets. Pick the entry point that matches where you are, or run the whole pipeline end to end.
- Loop
“We need a working robot.”
You're a manufacturer or enterprise with a specific task to automate. Loop takes you the whole way — plan the task, collect and curate data, transform it, train a policy, and deploy it on your hardware.
- Loop · collect
“We need data collected for our robot.”
You have a target robot but no operation to scale data collection. Run collect on Config's Data Studio with your hardware, or deploy on-prem to your site — and use only that module.
- Loop · transform
“We have the operation, but need conversion.”
You already run on-site data collection — often where data can't leave the building. Run transform on-prem to convert human demonstrations to your embodiment and build training-ready datasets. Same platform, fully private.
- Catalog
“We just need ready-made datasets.”
You're building a model or running research and need pretraining-scale robot data without operating any pipeline. Order from Catalog by task, embodiment, and volume.
Talk to our team.
Tell us about your task, your robot, or the data you need. We’ll show you how Config fits.