How Iterel works

Tasks are the unit of work

You don't chat with Iterel — you create tasks. Every message is triaged: a question becomes a Chat (a cited, saved answer, no work performed); a request to build becomes a Task (status, owner, receipt). A submission never produces both. This routing is automatic and invisible — there's no "Conductor" place to visit, just the composer and the results.

The stage shows one artifact with many faces

Your product is a single thing. The Design canvas, the Architecture map, the Live preview, and each task's Changes are all views of it. Edit a frame and the code behind it updates; a new page created by a task shows up on the map automatically. You switch faces; the artifact stays one.

Context makes agents smart

A project accumulates knowledge: uploaded docs, imported Jira/Confluence/Notion content, research outputs, and the decisions recorded in your threads. Every agent reads it, so generated copy is audience-correct instead of lorem ipsum. Ask the context anything and get a cited answer.

The shell that holds it together

Work lives in the left panel. Making happens in one of two modes — Build (the main studio) and Design (the design tool). Architecture isn't a mode; it's a full-bleed section you toggle on with the map control (or the M shortcut) and off to return exactly where you were. Everything converges on one entry point: the composer.