Use cases

AI diagrams for software architecture

Architecture diagrams die as screenshots in slide decks. Draw them as living boards instead — stencils, real connectors, version history — so the diagram survives the next refactor.

How you'd draw it in noddle

The editor underneath is built for exactly this: structured shapes and connectors that behave like Lucidchart, not free-floating lines.

  • Build from the AWS, Azure, GCP, and Network stencil sets, and wrap service boundaries in dashed containers.
  • Connectors are orthogonal and auto-routed with draggable waypoints — move the database and every arrow re-routes itself.
  • Set the request path edges to a beam flow to show how a request travels; export it as a looping GIF for the design-review PR.
  • Use pages as zoom levels, C4-style: system context on page one, one page per service below it.
  • Got years of diagrams in draw.io? Import the files and keep editing — then export SVG for docs whenever you need a still.

What the AI co-editor ✦ does

The AI co-editor ✦ is good at the first 80% of a diagram — the part that is typing, not thinking.

  • Describe the system in a sentence or two and it drafts the boxes and arrows; paste Mermaid from your docs and it becomes an editable board.
  • Iterate by asking: “add a read replica and a cache in front of Postgres” — Claude draws it live, with its own ✦ cursor, merging onto your latest edits.
  • Right-click any part of the board and “enrich with AI” to fill in the layer you sketched too thin.

What it will not do: read your Terraform state or discover your infrastructure. You — or the AI, from your description — draw what you know. The win is that the result stays editable and versioned, so drift is a drag away, not a redraw.

Start with a board, not a blank page

Free plan, no credit card — draw it by hand, from a prompt, or from a whiteboard photo.