Frequently asked questions.
Short answers to the questions we hear most. If yours isn't here, get in touch.
What is Agentuor?
A unified data infrastructure platform for building, evaluating, and deploying Physical AI and embodied robotics systems. It spans data collection, multimodal annotation, quality and review, model evaluation, and agent intelligence.
What does "agentic feedback" mean here?
Software agents analyze annotation work as it happens, incorporate reviewer input, and recommend labels and workflow actions. Each recommendation carries a confidence score and provenance, and a human must approve it.
Do agents ever change my data without approval?
No. Recommendations are never applied automatically. Every change to the dataset is confirmed by a person and logged.
Which data types are supported?
2D images, 3D data, LiDAR point clouds, and video, plus synchronized telemetry such as joint states and force-torque readings.
What kinds of robots is it used for?
Mobile manipulators, humanoids and legged robots, warehouse and logistics robots, autonomous mobile robots, and teleoperation programs, among others.
What is the difference between SaaS, managed, and hybrid?
SaaS gives your team full control inside your own workspace. Managed services means we execute end to end with our workforce and experts. Hybrid mixes the two — for example, keeping sensitive tasks in-house and bursting volume to us.
Can I import existing datasets?
Yes, with their labels and history mapped to the Agentuor schema.
Where is my data stored?
In the region you choose, or in your own storage under a hybrid deployment. See the Security page.
How does model evaluation work?
You bring policy outputs or execution logs; Agentuor scores them against ground truth, sliced by real-world conditions, and links failures back to the source data.
How do I get started?
Request a demo and we'll walk through the platform on your own data or a sample robotics dataset.