AI Studio

Hanzo AI Studio is a visual workspace for building AI apps and pipelines — wire models, tools, and data into a node graph, test it live, and ship it as an API. No boilerplate, no glue code.

Build Visually

Compose a workflow on an infinite canvas: drop in LLM, vision, and audio nodes, connect Marketplace models, add MCP tools and function-calling steps, and branch on outputs. Studio supports MCP servers for every function-calling-capable model hosted on Hanzo, so agents can use real tools inside a graph.

Test & Iterate

Run the graph directly in the canvas and inspect every node's input and output as data flows through. Swap a model, tweak a prompt, and re-run — iteration is immediate, and the models are the same ones you can call through the LLM Gateway.

Ship as an API

Publish a workflow and Studio serves it behind your Hanzo API key, so you can invoke it from anywhere with a standard request:

curl -X POST https://api.hanzo.ai/v1/studio/runs \
  -H "Authorization: Bearer hk-..." \
  -H "Content-Type: application/json" \
  -d '{
    "workflow": "my-workflow",
    "input": { "prompt": "Summarize this page" }
  }'

The exact route for a published workflow is shown when you publish it; see the API Reference for the full surface. Studio apps authenticate with the same hk- keys as the rest of the platform — see API Keys.

Related

• Marketplace — models to drop into a graph
• Crawl — feed web content into a workflow
• LLM Gateway · MCP · Chat