Your First Question

Run your first deliberation end to end — pick a mode, write a prompt worth deliberating, and read the result properly.

The fastest way to understand Dr Moot is to run one real question through it. Pick something you actually need a second opinion on — a decision, a plan, a piece of text you're unsure about.

1. Start a new chat

Open the Lab and click New chat in the sidebar. Each conversation carries its context forward, so a clean question deserves a clean chat.

2. Pick a mode (or don't)

The mode picker sits in the composer toolbar. Auto is the default and a good first choice: a router model reads your question and picks the right deliberation shape for it.

Auto mode routing between Relay and Moot

Choose manually when you already know what you want:

You want	Pick
A fast second opinion on a draft or decision	Relay
Two models genuinely arguing opposite sides	Debate
The deepest review, with a chaired verdict	Moot

Plan note

On the Light plan, runs use Relay regardless of what's requested — Debate and Moot are Pro features. Nothing errors; the run simply downgrades. See plans & limits.

3. Add context if the question needs it

Paperclip — attach the document the question is about (up to 5 files, 10 MB each). Wait for uploads to finish before sending.
Globe — turn on web search when the answer depends on anything that changed after the models' training data: prices, releases, news, live policy.
Rounds — in Debate or Moot, raise rounds to 2–3 for contested questions. Start with 1 for anything simple.

4. Write a prompt worth deliberating

The panel can only argue about what you give it. Strong prompts state:

the decision or problem,
the constraints,
what a good answer should optimize for,
the output you want back.

I need to decide whether to migrate this workflow now or wait one quarter.
Compare the operational risk, product upside, and team cost.
Give me a recommendation and the strongest case against it.

Asking for "the strongest case against it" plays to Dr Moot's structure — the Sceptic seat exists to build exactly that.

5. Watch the deliberation

Steps stream live as models write. You'll see the stages fill in — openings, critiques, revisions, and (in Moot) the Chair's verdict. A few things worth knowing while it runs:

The send button becomes Stop. Stopping keeps everything produced so far.
Refreshing the page reattaches to the in-flight run — you won't lose it.
In Auto, a badge shows the routing decision (e.g. Auto → Moot).

6. Read the result in two layers

The final answer is what the panel wants you to act on. The deliberation above it is why you should — or shouldn't.

Before acting on an answer that matters:

Read the Sceptic's critique (or Model B's, in Relay). Did the final answer actually address it?
Check the confidence level on Moot verdicts — High, Medium, or Low — and any dissent notes. Dissent means the panel didn't fully agree; slow down.
If web search ran, open the numbered sources and confirm they say what the answer claims.

7. Follow up or reset

Stay in the same chat when the next question builds on this one — prior context carries forward.
Start a new chat when you want the panel to come at a question fresh.

Modes — choose the deliberation shape on purpose.
Results & history — everything a finished run can tell you.