Yes — drawing and exporting in your browser is free and needs no account. Sign in only when you want to keep the diagram in Windy Docs, versioned and shared with your agents.

Do I need to install anything?

No. It runs in the browser, no login, and works offline.

What diagrams can I make?

System architecture, flowcharts, activity, state and finite-state-machine diagrams, entity-relationship (ER), class, and screen-flow diagrams.

Why not just describe my architecture in the prompt?

Because you think faster in pictures and your agent reads text more precisely. Drawing lets you reason visually; exporting hands the agent an exact, structured description instead of prose it has to interpret.

How does my agent get the diagram?

Export it to the AI-native text format and hand it over directly, or sign in and save it into a Windy project — your agents then read it over a project-scoped MCP endpoint, as part of the design they build from.

Free tool · no login

Draw your system visually. Your agent reads it as text.

You think best in pictures; coding agents think best in text. Draw system architecture, flows, state machines, ER and class diagrams in your browser — free, no login — then export to an AI-native format your agent reads precisely. Sign in and the same diagram becomes a first-class document inside Windy Docs.

Try it in your browser Start free

schema.er · canvas✎ drawn by hand

export → AI-native

schema.er.json✓ agent-readable

{
  "type": "er_diagram",
  "entities": ["Org", "User", "Subscription", "Project"],
  "relationships": [
    { "Org → Subscription": "exactly_one ↔ exactly_one" },
    { "Org → User": "zero_or_many ↔ zero_or_many" },
    { "User → Project": "zero_or_many ↔ zero_or_many" },
    { "Org → Project": "exactly_one ↔ zero_or_many" }
  ]
}

one schema — a picture for you, structured text for your agent.

Why it matters

Humans operate visually. Agents operate on text. This closes the gap.

Drawing is the fastest way for a person to think through structure — boxes, arrows, states, relationships. But a coding agent can't see your canvas; it reads text, and it reads it precisely. Windy lets you draw the way that's natural for you, then export the very same diagram into an AI-native textual format the agent understands exactly. One artifact, two audiences: a picture for you, structured text for the machine — no re-explaining the architecture in every prompt.

How it works

Draw it. Export it. Hand it to your agent.

Draw it visually

open the editor in your browser, no account needed; pick a diagram type and lay out boxes, arrows, and states by hand.

Export to an AI-native format

turn the drawing into structured text your coding agent reads exactly — give it the architecture instead of describing it in prose.

Sign in to keep it in Windy Docs

saved, the diagram becomes a first-class document type alongside your specs — versioned, and read by your agents over MCP.

Start drawing — it's free and there's nothing to install.

Try it in your browser Start free

windylogic.ai

Examples & templates

Start from a familiar diagram type.

System architectureFlowchartActivity diagramState diagramFinite state machineEntity-relationship (ER)Class diagramScreen flow

FAQ

Questions, answered.

Keep exploring.

AWS architecture diagrams How it works