5 Patents | 9,500+ API Endpoints | 18 Native DLLs | The AI-Native Spatial Runtime — Now in Early Access Explore the runtime →

Proof of the platform

The demo is the platform.

Capture your room. Play it. Your space is the game. RAKU HOOPS isn't a demo video — it's a live pipeline you run yourself at rakuai.com: scan your own room, then shoot hoops against your actual furniture.

Every stage you play through is a real, shipped platform component — capture, on-device rendering, scene understanding, physics, an MCP tool surface, and a bit-exact native core. Play it, then build on the exact same parts.

Play the proof → See how it maps ↓

Marquee milestones

NVIDIA Inception member AWS Activate approved — $10K credits Live MCP server at mcp.rakuai.com — versioned 17-tool contract RAKU HOOPS live at rakuai.com — play the proof Your room is the level — real physics on your real furniture On-device WebGPU splat rendering — no server video stream MCP scene-object read tool merged — full agent control shipping next RakuSplat contract v1 — browser JS & native C++ proven bit-exact
1

Capture

Scan a room with your phone.

What you saw
You pointed a phone camera at a room and captured it. No app install, no depth sensor, no rig — it runs in the browser.
The component
Raku Capture — a Progressive Web App that does the capture-side camera and frame work on your device. Open it directly at rakuai.com/capture-app/.
Build on it
Turn any camera into an input to a world model — the phone in your pocket today, smart glasses and robots on the same runtime next.
2

Reconstruct

The room comes back as a 3D scene — drawn on your device.

What you saw
Your capture reappears as a 3D Gaussian-splat scene you can orbit and inspect, rendered live in the browser.
The component
On-device splat rendering. WebGPU by default, WebGL2 fallback — the client tier is compute. No video is streamed from a server; the pixels are drawn on your own GPU.
Build on it
Ship spatial scenes to any WebGPU / WebGL2 device without a render farm or a per-frame streaming bill.
3

Understand

The scene is parsed into objects.

What you saw
The splat resolves into discrete objects. The in-game HUD reads it back honestly:
OBJECT COLLISION · N cells · N obj
The component
HP-6 object collision. Voxel occupancy is built from the splat centers, then clustered into per-object 3D bounding boxes. The HUD row only appears after a real object map is built — there is no fake object collision.
Build on it
Ask a scene what's in it: occupancy grids and per-object bounding boxes an app or an agent can query.
4

Simulate

Real physics against your room.

What you saw
The ball bounces off the actual furniture you captured. This isn't a canned level — the level is your capture.
The component
Physics runs against the object map derived from your scene, using one versioned collision model — the RakuSplat contract v1.
Build on it
Games, training simulations, and robot planning that run against real captured spaces instead of hand-authored levels.
5

Drive with AI

Any LLM can drive the world over MCP.

What you saw
The world is drivable by an AI agent — MCP-first and BYO-LLM. An agent connects its own model to mcp.rakuai.com and acts through tools, not a bespoke integration.
The component
Live MCP server. A versioned 17-tool native contract (5 read-only + 12 mutation); 154 tools registered in all. New: get_scene_objects lets an agent ask "what objects are in this room?" and get per-object world-space bounding boxes back — with an honest empty result when no object map exists.
Build on it
Claude, ChatGPT, Gemini, Copilot — bring your own model and drive the same spatial tools. No server-side LLM, vendor-neutral by design.

Where MCP control stands, honestly: the MCP server is live and the get_scene_objects tool is merged — an agent can query a scene's objects today, and gets an honest empty result when no object map has been published. The full capture→agent loop — your browser publishing the object map so an AI reads back the real furniture you just scanned — is landing next, not live. We will not claim it as proven until a recorded round-trip against a fresh capture exists. Fake success is forbidden here.

6

Native core

The same math, browser and native — proven identical.

What you saw
The collision math you just played runs in the browser — and the exact same contract runs in a native C++ core.
The component
RakuSplat contract v1 — one JSON-Schema contract, three tiers. The browser JS reference and a native C++/Linux core (libraku_splat.so) are proven bit-exact by a 74-check conformance suite (integer outputs exact, floats within 1e-6). The header is the single source of truth, ABI drift-gated.
Build on it
Pin to one versioned schema and get identical spatial semantics from browser to native — port a tier without silent divergence.

Where the native core stands, honestly: the RakuSplat v1 contract, the browser JS reference, and the native C++/Linux core are shipped in the open repos, and the conformance suite proves they agree bit-for-bit. Native activation is rolling out tier by tier (Rung 1) — we are not claiming production endpoints are native-served yet. When a stage isn't ready, it says so and fails loudly. Fake success is forbidden here.

Stop watching demos. Run the pipeline.

Smart-glasses OEMs, LLM labs, enterprise teams, and startup programs: the fastest way to evaluate RakuAI is to play the proof. Every stage you touch is a component you can build on.

Try it live → Connect your AI over MCP → Talk to us →