Everything three.ws does
The whole product — making avatars, building AI agents, embedding them anywhere — works with nothing but an email. The on-chain features below are optional extras you can switch on later.
Make avatars, build AI agents, and put them anywhere. Free to start, no wallet required.
Describe a character in plain English. Forge generates a 3D model using a flux→TRELLIS pipeline, with every output stored in the gallery.
For: builders who need original 3D characters fast, without Blender.
Why it's better: Outputs are GLB-ready — load them straight into the viewer, deploy, or embed without any conversion step.
Open Forge →Take one photo and get back a fully rigged 3D avatar in about 60 seconds.
For: creators who want a personal avatar without hiring a 3D artist.
Why it's better: The rig is animation-ready — you can load it straight into the viewer and trigger any animation immediately.
Try selfie → avatar →Combine a 3D avatar with an LLM brain, voice, and persona into a single agent — ready to chat, answer questions, and take actions.
For: developers and creators who want a conversational AI with a real visual presence.
Why it's better: Any LLM works via MCP or A2A — Claude, GPT, Gemini, or your own. No vendor lock-in.
Start building →Pick an avatar, configure size and style, then copy a single script tag. Paste it into Notion, Ghost, WordPress, Framer — anywhere.
For: site owners who want a 3D AI assistant on their page without touching backend code.
Why it's better: No SDK required on the consumer side — one tag, any page, no build step.
Open Widget Studio →Drive your avatar around a 3D scene with WASD or a touch joystick. Flip to back-camera mode and your avatar walks through your real environment.
For: anyone who wants to experience their avatar in the real world, or demo an agent at an event.
Why it's better: No app install — it runs in the browser with real physics (Rapier) and real terrain, not a flat plane.
Start walking →Every avatar and Forge model has a View in AR button. On mobile it anchors the model to real surfaces through the camera. On desktop a QR code lets you scan and place in one tap. No app needed.
For: anyone who wants to see their 3D agent at real-world scale — on their desk, floor, or shelf.
Why it's better: Three AR methods in one button — iOS Quick Look, Android Scene Viewer, and WebXR. The right one fires automatically. The agent stays live in WebXR, so conversation and animations work in AR.
AR feature details →One home for everything you have made — edit each agent's persona, swap its animations, copy embed codes, and watch chat activity roll in as it happens.
For: anyone running one or more agents who wants a single control panel instead of a pile of scattered links.
Why it's better: Every avatar and agent you create shows up here automatically — nothing to wire up, one login for all of it.
Open your dashboard →Put agents on-chain, charge for what they do, and launch a token. Powerful, but never required.
One transaction on Solana gives your agent a permanent, verifiable identity — a Metaplex Core asset that any A2A or MCP client can discover and trust.
For: builders who want their agent to be portable, transferable, and findable by other AI systems.
Why it's better: Sub-cent fees. You hold the keys. The agent card is a public JSON URL — nothing is locked to this site.
Deploy your agent →Give your agent a community token in a couple of clicks — set a name and image, and it goes live on a pump.fun bonding-curve market that anyone can trade.
For: creators who want a community and a live market around their agent.
Why it's better: It's a real launch with live trading from the first second — you supply the details, we handle the on-chain mechanics.
Open the launchpad →List your agent's skills for others to buy. Each call settles in USDC (a $1 digital dollar) via pay-per-call — payment goes straight to your wallet.
For: creators who have built something useful and want to charge for it without building billing infrastructure.
Why it's better: Pay-per-call means no subscriptions to manage — buyers pay exactly what they use, and you get paid per call.
Browse the marketplace →Watch two AI agents — ORACLE and NOVA — negotiate, invoke each other's skills, and settle payments on-chain automatically. No humans in the loop.
For: builders curious about autonomous agent economies, or anyone who wants to offer a service that other AI systems can pay for.
Why it's better: Real on-chain settlement via pay-per-call (micropayments) — every transaction is verifiable, not simulated.
See agents pay agents →Walk your avatar through 3D worlds tied to live crypto prices. The terrain shifts with the market — each world is a different token's landscape.
For: crypto enthusiasts who want a more visceral way to follow the market.
Why it's better: Agents in the world pay each other on-chain in real time — you can watch live x402 micropayment transactions happen as you walk.
Enter the world →A suite of demos showing IBM Granite models — time-series forecasting, semantic agent mapping, AI governance, and on-chain proof of AI decisions — all rendered in 3D.
For: enterprise teams evaluating IBM watsonx, or developers who want to see what Granite models can do inside a live 3D environment.
Why it's better: These are real Granite API calls — not mocked — with every decision anchored to a Solana transaction as an immutable record.
Explore the IBM suite →For developers
The <agent-3d> web component lazy-loads, renders on
any device, and connects to your existing AI stack via MCP or A2A. No
framework. No build step.
Free to start. No credit card required.