Results and History
How to watch a live run, read verdicts and dissent, verify sources, and manage your saved conversations.
Dr Moot's answer is never just the last paragraph. Every run produces a final answer and the deliberation record that justifies it — and both stream live while the run happens.
While the run is live
- Steps stream as they're written. You watch the drafts, critiques, and the final answer appear token by token — including which stage the panel is currently in.
- Stop is always available. The send button becomes a stop button during a run. Stopping keeps everything produced so far in the transcript.
- Refresh-proof. If you reload the page (or your connection drops) mid-run, the run keeps going on the server and the Lab reattaches and catches you up.
- Errors are explained. If a run fails — a model outage, a timeout, a budget limit — you get a specific, human-readable reason, and each step is retried once before giving up.
Reading a finished run
The deliberation stages
Above the final answer, the run is organized into stages you can expand or collapse: openings, critique rounds, revisions, web search, verification, and the verdict. While a run is live the active stage is open; afterwards it collapses into a compact summary you can reopen anytime.
What to look for, in order of value:
- The critiques. This is where the panel earns its cost. Did the Sceptic find something real? Did the revision actually address it?
- The verdict (Moot) — see below.
- The evidence — search results and verification, if they ran.
Verdicts, confidence, and dissent
Moot runs end with a Chair verdict:
- A confidence level —
High,Medium, orLow. TreatLowas "the panel is telling you it isn't sure." - Dissent notes when the panel didn't fully agree. Dissent is a feature: it tells you precisely which part of the answer is contested, so you know where to apply your own judgement.
Auto routing badges
Runs in Auto mode show the routing decision — e.g. Auto → Relay or Auto → Moot — so you always know what you paid for. Questions the router flagged as high-stakes also run a verification pass over the panel's factual claims, and are badged accordingly.
Sources
When web search contributed, sources appear as a numbered list with links. Before acting on a claim, click through: confirm the source exists, says what's claimed, and is current enough to matter.
Follow-up questions
Conversations carry context forward: a follow-up in the same chat sees the earlier questions and answers.
- Stay in the chat when the next question builds on this one ("now draft the email announcing that decision").
- Start a new chat when you want fresh eyes — earlier context anchors the panel, which is sometimes exactly what you don't want.
Very long conversations are trimmed to the most recent turns when building the panel's context, so the freshest exchanges always win.
Managing history
The sidebar keeps every conversation:
- Reopen any conversation — the full deliberation record is preserved, not just the answers.
- New chat resets context.
- Delete conversations you no longer need from the sidebar.
Habit worth building
When an answer surprises you — pleasantly or not — don't re-ask. Expand the critiques first. Nine times out of ten the deliberation already contains the caveat, the counterexample, or the missing assumption you were about to go hunting for.
Attachments and Web Search
Give the panel your documents and images, or let it gather live evidence — file types, limits, extraction behavior, and error messages explained.
Settings
A map of the settings overlay — account, security, notifications, billing, organization, and where the Dr Moot–specific preferences live.

