Skip to main content
The CTO is ADE’s persistent, project-level operator: one identity per project, not a rotating set of chats. While a lane chat works inside a single worktree, the CTO sees the whole project — every lane, PR, and connected Linear issue — and reasons across them. It is a single long-living chat thread with a memory that carries decisions and context forward across sessions, context compaction, and model switches. Reach for the CTO when a question or task spans more than one lane, needs planning, or benefits from remembering earlier decisions.

​
What the CTO is

Persistent identity

One CTO per project, with continuity that survives long conversations — not a fresh chat each time. Its voice is fixed by ADE, so it reads the same on day one and day ninety.

Project-wide context

It reasons across every lane, PR, and connected Linear issue, instead of one worktree at a time.

Durable memory

It saves the decisions, preferences, and standing facts that matter, and recalls them in later sessions — even after you switch its model.

Reads Linear

When Linear is connected, the CTO can browse issues, read comments, and post lightweight updates without leaving the thread.

Knows ADE itself

It knows the application, not only your repository. Ask where a setting lives or why a control is disabled and it answers from ADE’s own documentation, with a deeplink to the place.

​
CTO vs. lane chat

The CTO doesn’t replace lane chats — it sits above them. A lane chat edits files in one worktree; the CTO holds the project-wide picture, plans the work, and remembers what was decided. Pick whichever matches the altitude of your question.
The CTO lives in its own CTO tab: a single chat thread plus a Settings sheet. There are no tabs to manage — open it and start talking.

​
Memory that carries over

The CTO keeps durable memory as part of the project, so it doesn’t start from zero each time. It records the decisions and preferences worth keeping, and ADE re-injects the relevant parts into every session — including after a long conversation is compacted or after you change the model mid-thread. You can see and edit what it remembers from CTO → Settings → Memory.
Ask the CTO to remember something explicitly — “remember that we deploy from main only” — and it will save it as a durable fact you can review later.

​
Switch models without losing the thread

You can change the CTO’s model at any point from Settings → Model. The thread, its transcript, and its memory carry over to the new model — a switch is not a reset. Favor a capable model for project-wide reasoning; drop to a faster one for quick lookups. The list is narrower than elsewhere in ADE on purpose: the CTO only runs on models that can take a message into a turn already underway, because it is interrupted constantly by the chats it starts and by its own wake-ups.

Set up the CTO

Pick a model, then optionally tune reasoning effort and connect Linear.

Linear

Connect Linear so the CTO can browse and update issues.