AI Chat

Chat Interface

Describe what you want to build and the AI generates a full project — contracts, components, hooks, and API routes — deployed to a live sandbox preview.

How the AI Chat Works

You type a message describing what you want. The AI processes it through a streaming pipeline via Server-Sent Events (SSE), and you see real-time progress in the UI as each step completes.

Text Messages

Your instructions and the AI's responses. The conversation keeps full context of every previous message in the project.

Generation Messages

Show streaming progress with pipeline steps, generated artifacts (files), and the current file being written in real time.

Clarification Messages

When the AI needs more info, it asks multiple-choice questions — for example, "Which chain do you want?" with selectable options.

The 6-Step Generation Pipeline

Every generation request flows through these six stages. You can watch each one complete in real time.

INTENT

The AI understands your request and identifies what you are asking for.

PLAN

The AI designs the project architecture, selects templates, and determines which feature flags to activate.

GENERATE

Code generation begins — smart contracts, React components, hooks, and API routes are created.

FIX

The AI automatically detects and corrects any errors found in the generated code.

VALIDATE

Validation checks run against the generated output to ensure correctness.

DEPLOY

The finished project is deployed to a sandbox preview you can open immediately.

What You Can Ask the AI

The chat handles everything from full project generation to targeted modifications on existing code.

Generate Full Projects

• "Create an NFT marketplace on Base with public minting"
• "Add a DAO governance dashboard" — generates Governor + Timelock + Token
• "Build a Solana token dashboard with analytics"

Add Features

• "Add a swap widget using Jupiter for Solana"
• "Add a public mint page with a countdown timer"
• "Include an analytics section showing holders and supply"

Modify Existing Code

• "Change the mint price to 0.1 ETH"
• "Switch the chain from Ethereum to Base Sepolia"
• "Update the token supply to 50 million"

Deploy Contracts

• "Deploy my token to Base Sepolia"
• "Deploy the NFT contract to mainnet"
• Triggers the full contract deployment flow

Feature Flag Activation

The AI follows a minimal inclusion principle — it only activates the features your project actually needs. No unnecessary dependencies.

How Flags Work

• Asking for "a landing page" — no blockchain providers, no Web3 dependencies are included.
• Asking for "an NFT mint page" — activates evm + evm-nft + mint flags.
• Asking for "a Solana token dashboard" — activates solana + analytics flags.

Example Conversations

Real examples of how the chat generates projects end to end.

Token + Dashboard

User:

"I want to create a token called ALPHA with 10M supply on Base, with a dashboard showing supply and holders"

AI Pipeline:

Flags activated: evm deploy analytics

Generated: ERC-20 contract, TokenDashboard component, deploy flow, API routes for analytics

NFT Collection

User:

"Create an NFT collection with 1000 items, 0.05 ETH mint price, and a public mint page"

AI Pipeline:

Flags activated: evm evm-nft mint

Generated: ERC-721 contract, PublicMintPage, NFTGallery, deploy flow

AI Asks for Clarification

User:

"Build me a swap app"

AI Response:

"Which ecosystem would you like?"

Solana (Jupiter)EVM (Uniswap)Both

Credit System

Each generation consumes credits tied to your subscription plan.

How Credits Work

• Each generation request consumes credits from your plan.
• If you run out of credits, the API returns a 402 error.
• You can top up your credits in the Billing settings page.

Tips for Effective Prompts

Get better results by following these guidelines.

Be Specific

Mention the chain, token standard, and features you want. The more detail you provide, the fewer clarification rounds you need.

Let the AI Ask

If your prompt is ambiguous, the AI will ask clarifying multiple-choice questions before generating. This is normal and improves the output.

Iterate with Follow-ups

Send follow-up messages to modify or add features. Each message in a project conversation has full context of all previous messages.

One Thing at a Time

For complex projects, start with the core request and add features incrementally through follow-up messages.

Ready to build with AI?

Describe your Web3 project and watch the AI generate it in real time.

Start Building