three.ws · deck

three.ws · Investor overview

Give your
AI a body.

three.ws turns any AI into a real-time 3D character — a face, a voice, and motion — that you can embed on any website in one line of code.

Try it live →
Live · drag to orbit

01 · The problem

AI got smart.
It's still invisible.

Every product now ships an AI. Almost all of them are a text box. The moment you want that intelligence to feel like a presence — a greeter, a host, a character, a teammate — you hit a wall.

Building a real-time 3D avatar today means assembling all of this:

  • A 3D artist to model and texture a character
  • A rigging pipeline so it can actually move
  • An animation system for idle, gesture, and lip-sync
  • A game engine and real-time render stack
  • An ML layer to connect a brain to the face

Months of specialist work. Embodiment is locked behind expertise almost no team has.

02 · Why now

Three curves just crossed.

1

Conversational AI went mainstream

Every product wants a face for the assistant it already shipped. Demand exists — the supply of embodiment doesn't.

2

Real-time 3D runs in every browser

WebGL and WebGPU put console-grade rendering on a plain web page. No install, no app store, no plugin.

3

Generative 3D crossed the quality bar

A photo or a single sentence now becomes a rigged, animatable character — in minutes, not weeks.

What's missing is the layer that connects them — and turns a model plus a personality into an embodied agent you can drop anywhere.

That layer is three.ws.

03 · The solution

One platform. Idea → embodied agent in minutes.

three.ws is the connective tissue between an AI model and a body the world can see and talk to. Four steps, no specialists.

Create

A character, fast

A selfie, a photo, or a text prompt becomes a rigged 3D avatar.

Think

Plug in a brain

Pick the model — Claude, GPT, Granite — and give it knowledge and a personality.

Move

Make it alive

Lip-sync to its voice, emotes, full-body motion capture, idle presence.

Embed

Ship anywhere

Two lines of HTML drop the agent onto any website, stream, or app.

No 3D modelling. No game engine. No pipeline to maintain.

04 · The product, in a sentence

Everything ships behind one line.

The avatar, the brain, the voice, and the motion all travel inside a single embeddable element. Paste it, and a living agent appears.

< 2lines of HTML to embed a full agent
0installs, plugins, or app-store steps
glTFopen 3D standard — portable everywhere
<!-- drop a 3D AI agent on any page -->
<script type="module"
  src="https://three.ws/agent-3d/latest/agent-3d.js"></script>

<agent-3d
  avatar="your-character"
  brain="claude"
  voice="on"
></agent-3d>

That snippet is the whole integration. The hard parts live on our side.

05 · Live demo

This is live.
Right now. In this slide.

The character beside this text is rendering in your browser from a standard 3D file — the exact same thing that ships inside the one-line embed. Drag it. Spin it. It's real.

  • Rendered client-side — no video stream, no server round-trip
  • Rigged and animated, ready for lip-sync and gesture
  • Portable glTF — works in any modern browser and engine
Live in-browser render · drag to orbit

06 · The product

Not a demo. A shipped platform.

Every block below is built, wired, and in production today — and they all connect to the same agent you create.

Avatar Creator

Selfie, photo, sculpt, or text prompt → a rigged, animation-ready 3D character.

Animation Studio

Idle, emotes, lip-sync, and full-body motion capture from a webcam.

Embeddable Widget

The one-line element that drops a talking agent onto any website.

3D Worlds

A space your agent and your audience can walk into and share.

Scene Studio

A browser 3D editor to compose, light, and export full scenes.

Developer SDKs & Toolslive on npm

Open packages and an agent tool server so any product can build on the stack.

07 · The market

We sit where three large markets converge.

Embodiment isn't a niche — it's the missing interface layer across three of software's fastest-growing categories.

Conversational AI

The brains, multiplying

Every app is adding an assistant. None of them have a face. We are the body for all of them.

3D & avatars

Content, democratized

Avatars, virtual production, and digital humans — once studio-only, now a prompt away.

Creators & CX

Presence, monetized

Streamers, brands, and support teams all want a recognizable character that's always on.

Each category is measured in tens of billions and growing double digits. The embodiment layer is the connective tissue across all three — and it's currently unowned.

08 · Who it's for

Same platform. Very different jobs.

One creation flow feeds four distinct, paying audiences — each already reachable through a working product surface today.

Businesses

Support avatars

A branded 3D greeter that answers customers 24/7 — a friendlier face than a chat bubble.

Creators

Stream characters

A lip-syncing on-camera avatar for video and live, without a pro VTuber rig.

Communities

Mascots & worlds

A character that represents a community — and a 3D space for it to gather in.

Builders

Personal agents

An assistant you can look at and talk to — and call from the API in your own product.

09 · Business model

Free to start.
Revenue scales with use.

Creation and embedding are free — that's the top of a self-serve funnel. Money follows the moment an agent does real work.

1
Create & embed — freeTurns every user and every embed into distribution.
Top of funnel
2
Premium generationHigher-fidelity avatars, faster pipelines, more capacity.
Usage-based
3
Developer plansSDK and API access, higher limits, and analytics.
Subscription
4
Team & enterpriseBranded embedding, support, and custom characters at scale.
Contract

10 · Distribution & moat

Built to spread.

The product is the marketing. Every integration is a billboard, and every developer who builds on us deepens the lock-in.

Open SDKs on npm

Published packages mean developers adopt us inside their own products — and stay.

Agent-native tools

Our capabilities are exposed as tools other AI agents can call — distribution where AI already lives.

One-line embed = reach

Every site that adds an agent becomes a live, branded demo seen by its whole audience.

Compounding asset library

Every avatar, animation, and world created makes the next creation faster and richer.

The full stack, integrated

Rivals ship a piece; we ship create-think-move-embed as one path no one else has end to end.

Standards-based

Open glTF output means no walled garden — which is exactly why teams trust building on it.

11 · Traction

What's already shipped.

While the category is still being named, we've shipped the whole path — in production, in the browser, today.

End‑to‑endselfie → talking, embedded agent works now
In‑browserruns on every modern device, no install
  • Dozens of product surfaces live in production — creator, studio, worlds, widget, and more.
  • Developer SDKs published on npm for outside teams to build on.
  • Agent tools live on a public model-context registry — callable by other AI.
  • One-line embed proven across real third-party sites.
  • Open glTF pipeline — created characters export and run anywhere.

We didn't write a roadmap to embodiment. We shipped it, then started compounding.

12 · Roadmap

From shipped to inevitable.

Now
  • Harden create → embed for every browser and device
  • Deepen real-time lip-sync and expression quality
  • Grow the SDK and developer surface
Next
  • Live voice + video conversation in the embed
  • A marketplace of characters, animations, and worlds
  • Team workspaces and branded enterprise embedding
Later
  • Mobile and AR placement of agents in real space
  • Multi-agent worlds and shared social presence
  • The default way any AI gets a body

13 · The ask

Give your AI a body.

The brains are everywhere. The bodies aren't built yet. We're the team that already shipped the layer that gives them one — and we're raising to make it the default.

three.ws three.ws @trythreews Let's talk.
1 / 14