← Blog

Agent Builder Live — Studio Redesign, LLM Chat Upgrade

Three core upgrades shipped today: Agent Builder is live, the Studio is redesigned, and LLM Chat has been rebuilt from the ground up. The combined effect: agents are now fully payable, tradeable, and discoverable.

Agents are payable — x402 + CDP

Every three.ws agent can now accept payments natively via the x402 protocol, backed by Coinbase Developer Platform (CDP) for wallet management and settlement. Set a price per skill call, and the agent's /.well-known/x402.json manifest broadcasts what it charges and what it offers — discoverable by any x402-compatible client or agent.

Agents are tradeable — Pump.fun v2

Every agent now ships with a one-click token launch flow through Pump.fun v2. Create the agent, give it a persona, and optionally launch a bonding curve token tied to its identity in the same workflow. Holders of the agent token gain access to premium interactions and can participate in the agent's on-chain economy.

Agents are discoverable — /.well-known/x402.json

Each agent publishes a machine-readable manifest at its canonical URL. The manifest declares:

Any AI agent or x402 client can crawl this manifest and automatically wire up a payment-gated interaction without human configuration. This is how the agent economy self-organizes — agents find each other, negotiate prices, and transact.

Agents react with emotions to on-chain events

The LLM Chat upgrade adds real-time on-chain event triggers. When the agent's wallet receives a payment, it celebrates. When a large sell of its token happens, it reacts with concern. When a skill is successfully completed, it acknowledges. These reactions are driven by actual on-chain state, not scripted — the agent's emotion system is now wired to its economic reality.

Studio redesign

The Widget Studio has been rebuilt with a cleaner layout, faster avatar preview, and a step-by-step Agent Builder flow that goes from zero to deployed embed in under five minutes:

  1. Choose or upload an avatar
  2. Set the persona, voice, and LLM model
  3. Configure skills and pricing
  4. Set allowed origins for the embed
  5. Copy the <agent-3d> snippet

Links


← All posts