Kindly Logo

Kindly

Sign in
The Data Backbone for Physical AI

Every robot skill needs training data.

The credentialed clinical data layer for surgical physical AI — capture, label, and lineage for the operating room and sterile processing. Built on the same pipeline that ingests demonstrations from simulation or hardware and tracks them from raw sensor data to trained VLAs and diffusion policies.

Start Labeling (Earn XP)Explore DatasetsAPI Docs
Projects Robots Label Data CLI MCP Compare
FoodforThought Hero
Clinical Vertical
In development — design partner program launching Q3 2026

The wedge: clinical-environment data for physical AI.

Generic robotics data is a commodity. The next unlock for foundation-model labs and surgical-robot OEMs is clinical-environment training data — captured with credentialed access, structured for VLAs and diffusion policies, and labeled to clinical workflow standards. We are building that corpus now.

Why us: Founding team includes a surgical RN and a hospital robotics coordinator. Credentialed clinical access is the differentiator — not the dataset hosting layer.

Explore the clinical data layer — live now

Published v0.1 4-layer annotation schema, machine-readable JSON Schema, two downloadable synthetic sample episodes, and an interactive episode viewer — explore the clinical corpus.

Honest status: no clinical data is live on FoodforThought today. The first design-partner cohort is being selected now; data and access programs ship over the next two quarters.

Join the design partner programSee the annotation schema

For foundation-model labs and surgical-robot OEMs evaluating clinical training data sources.

Built for researchers at leading labs

Berkeley RAILStanford IRISNYU RIPLCMU RIMIT CSAIL

How It Works

The complete data pipeline for Physical AI—from raw sensor data to deployed skills.

1

Capture

Record demonstrations from Artifex simulation or hardware via ATE CLI.
2

Ingest

Automatic ingestion from Bridge V2, Open X-Embodiment, or your own fleet.
3

Label

Community-powered annotation with XP rewards, streaks, and leaderboards.
4

Track

Full lineage from raw sensor data to trained model. Never lose provenance.
5

Deploy

Push skills to robots with cross-embodiment compatibility checking.
6

Iterate

Feed deployment results back into the pipeline. Every run improves the next generation of skills.

The Data Flywheel

Every robot execution generates training data. Every label improves the next model.

CaptureLabelTrainDeployIterate(repeat)

Built for Physical AI at Scale

Whether you're training VLAs, diffusion policies, or behavioral cloning models.

130K+ Trajectories

Curated from Open X-Embodiment and Bridge V2 datasets with consistent formatting.

22+ Robot Embodiments

Cross-compatibility analysis for skill transfer between robot platforms.

Community Labeling

Gamified annotation with XP, streaks, achievements, and global leaderboards.

API + CLI + MCP

Integrate with any workflow. Full programmatic access to the data platform.

CLI Tools

30+ commands for the complete robotics development workflow. Generate skills, check cross-robot compatibility, upload data for labeling, and integrate with your editor via MCP.

Skill Generation

Scaffold robot skills in 30 seconds from 5 built-in templates: pick & place, navigation, inspection, assembly, pouring.

Compatibility Scoring

Check if skills transfer between robot types. Kinematic, sensor, and compute scoring across 22+ embodiments.

Data Labeling

Upload demonstration videos for community labeling. Bounding boxes, action segmentation, polygon masks, Q&A.

MCP Integration

89 MCP tools for Cursor and VS Code. Build robot skills without leaving your editor.

See it in action

Terminal — FoodforThought CLI
Explore the CLI

Featured Datasets

Curated robotics datasets for training and evaluation

Featured Projects

Example projects demonstrating the data-to-skill workflow

Featured Robots

Popular robot platforms with datasets and skills

Community Contribution

Help Train the Next Generation of Robots

Label robotics data to earn XP, unlock achievements, and climb the leaderboard while contributing to the open robotics community.

Try Labeling — No AccountBrowse TasksLeaderboard

See it in action

Watch how easy it is to check skill compatibility and generate adaptation plans between different robots using our CLI.

Terminal — FoodforThought CLI

Try it yourself: Install the CLI

How FoodforThought is organized

FoodforThought is structured around projects and reusable content so you can move quickly from raw demonstrations to reusable skills.

Projects

Group artifacts, datasets, configs, and skills for a specific robot task.

Track progress across stages (raw data, processed, labeled, skills) with lineage.

Browse

Discover repositories, datasets, configs, and task recipes shared by others.

Filter by robot, domain, and stage to find content you can reuse.

Robots

Explore robot profiles, specs, and compatible content.

Jump into projects, demos, and configs that match your robot model.

Search & Contribute

Search across everything, or add new datasets and programs.

Use search when you know what you need; use Contribute to share back with the community.

Beyond the Platform

FoodforThought is part of the larger Kindly ecosystem. See how our technology is being deployed.

KindlyforUs

Robots Making a Difference

These datasets power our community cleanup missions. See robots in action cleaning beaches and building homes.

View Cleanup Missions

KindlyforYou

Scale Your Fleet

Ready to deploy these skills commercially? Our enterprise platform manages fleets with real-time monitoring.

Explore Fleet Management
© 2026 Kindly Robotics