Raven is an innovative platform that makes it easy to combine kinematic control, vision-language AI, and imitation learning — the three core technologies driving modern robotics. Start exploring on day one, go as deep as you want.
Free Download
Every Raven skill runs the same three-step loop — powered by real AI, on a real arm.
The arm's camera captures the scene. Raven uses your phone's camera or the arm's own — whichever gives the best view.
A Vision Language Model reasons about what it sees and decides what action to take — based on a prompt you wrote in plain English.
The arm executes the chosen motion — a recorded trajectory, a pick-and-place sequence, a drawn stroke, or a custom skill you built.
Motion Intelligence
Industrial robots are taught by capturing key positions — waypoints — and letting the controller generate smooth motion between them. Raven uses the same concept: jog the arm to a position, capture it, move to the next. The trajectory engine interpolates the path and replays it with consistent precision.
Jog the arm to each key position and tap to record. No programming — just physical demonstration.
Raven's cloud trajectory engine interpolates smooth, continuous motion between your waypoints — the same technique used in industrial robotics.
The trajectory is stored in your skill and replays with exact repeatability. Wire it to a VLM decision and the arm acts intelligently on demand.
Built-In Motion Templates
From Motion to Intelligence
When you replay a trajectory, Raven can simultaneously record joint states and camera frames as a synchronized episode — the same format used by research-grade robotics pipelines. Collect enough episodes and you have a dataset ready to train an ACT model or any other imitation learning architecture. Raven datasets are fully LeRobot compatible, so your data works seamlessly with the broader open-source robotics ecosystem.
Raven grows with you. The same arm and app take you from your first recorded motion to building and publishing AI-powered skills — no upgrade required, no platform switch.
Think of it as an app store for your robot arm. Browse skills created by the community, install them in one tap, and run them immediately. Or build your own and publish them for anyone with a Raven to use.
Browse a growing library of community-created skills — games, creative tools, AI experiments, useful automations, and more. Filter by category, difficulty, or what the arm does.
Install any skill with one tap. Every skill is a self-contained package — motions, AI prompts, and UI all bundled together. No configuration. Open the app and go.
Build something in the Skill Studio and share it with the world. Every skill you publish becomes available to every Raven owner — your creation, running on thousands of arms.
Skill categories
The store launches with Raven. Join the waitlist to be among the first to publish.
The Skill Studio is a no-code environment for creating, testing, and publishing AI-powered skills. Record motions, write prompts, design a UI, chain logic — then publish to the community with one tap.
Guide the arm through a sequence and capture it as a replayable trajectory — the foundation of every skill.
Tell the VLM what to look for and how to respond. The arm acts on whatever the model decides.
Package your skill as a portable file and publish it to the community. Anyone with a Raven arm can install and run it.
Supported Devices
Raven Studio works across all major platforms. Use a phone or tablet to control the arm wirelessly, or run the full desktop app on Mac or Windows for the Skill Studio and local AI models.
On-Device AI
Beyond cloud AI, Raven supports a growing set of on-device language models that run entirely on your device — no internet, no cloud credits consumed. Perfect for camp environments, travel, or anyone who values privacy.
Runs on virtually any device. Ideal for camp tablets and basic vision tasks with zero latency.
Google's compact multimodal model. Strong vision reasoning on modern phones and tablets.
Powerful vision-language reasoning for demanding skills. Best on modern flagship devices and desktops.
Access a broad range of frontier vision-language models via the cloud. The most capable option — uses cloud credits.
Raven supports up to three simultaneous camera sources — each assigned its own role. The more the AI can see, the smarter it acts. Use what you have: the device in your hand, an old phone on a stand, or a USB camera mounted overhead.
The primary camera. High quality, always available, and works the moment you open the app.
Give a retired phone a second life. Mount it overhead or at an angle for a second perspective — Raven sees it as a network camera over WiFi.
Connect any USB webcam to Mac or Windows. Great for fixed overhead shots, lab setups, or high-resolution scene capture.
The optional camera mounted on the arm itself gives a first-person view of exactly what the gripper sees as it moves.
Every Raven arm includes the app, Skill Studio, built-in skills, and cloud AI access — everything in one.
The complete Raven platform — fully assembled and ready to run. Skill Studio access included for every arm.
Multi-arm packages for summer camps, after-school programs, and schools. Includes group cloud credits, instructor support, and curriculum materials.
Raven is currently in limited availability. Get on the list and we'll reach out as soon as arms are ready to ship.
Thanks for your interest in Raven. We'll reach out to with early access details as soon as arms are ready to ship.