Learning · Community report
Run deep research on a livestream event
In a post on X, a Muse user describes connecting a research API and having Muse run deep research on two days of a livestream event, transcripts and community notes included. The promised follow-up is missing, so nothing is known about the report's accuracy; what follows is a prompt to adapt and what to check.
What you can get
What it is
Research report
A compiled research report drawing on event transcripts and community notes, covering the full date range asked for.
It came back right when
- The report covers the full date range you asked for.
- Sources used, whether your own notes or a connected tool, are identifiable.
- You reviewed the report before treating it as accurate.
- Nothing was published or shared from the report without you.
The checks are editorial, not from the source.
- Define the scope
Event name and the exact date range or sessions to cover.
- Provide what you have
Existing transcripts and community notes.
- Connect sourcing if needed
A research or search tool for additional material.
- Review the reportStops for your approval
Check against transcripts or your own knowledge of the event.
Editorial, not from the source
Works when
- You have a multi-day event with transcripts or notes to draw from.
- You're fine reviewing a compiled report rather than treating it as final.
- You can connect a research tool if deeper sourcing is needed.
Does not work when
- You need guaranteed completeness across every session of the event.
- You have no transcripts, notes, or tool access to point it at.
- You need the report fact-checked against primary sources before anything else happens.
Run deep research on a livestream event: review template
| Item | What it is based on | Status |
|---|---|---|
| [Event and date range] | [Your scope] | [Confirmed covered] |
| [Source material used] | [Your transcripts/notes plus any tool] | [Identifiable per claim] |
| [Compiled report] | [Research output] | [Spot-checked by you] |
The source case
On X, @ChrisUniverse reports connecting a research API to Muse and giving it a first job: full deep research on two days of a livestream event, including transcripts and community notes. The post promises a followup thread with what came back, but the report's actual content is not included in this text; checked on 2026-10-09, there is no detail here on the report's length, accuracy, or how long it took.
A public report by someone outside this library. Not independently reproduced here.
From the source
Source excerpt
First job I gave it: full deep research on the Day 1 & Day 2 @bot Galaxy livestreams — transcripts, community notes, the whole nine yards!
Quoted from the original posts. Not written by this guide, and not a result produced here.
Suggested prompt
Bring Event name and date range, Any transcripts or notes you already have. Acts Compiles a research report from transcripts, notes, and any connected search tool. Does not publish the report anywhere without you. Source community report, checked 2026-10-09.
Written by this guide as a starting point, not quoted from the source. Your edits stay in this browser.
Other ways to ask
Editorial rewrites of the brief above, for a different emphasis. Copy them separately.
What you need
Editorial, not from the source
You supply
- Event name and date range
- Any transcripts or notes you already have
Connected
- A research or search API, if more sources are needed
Approved by you
- None beyond compiling and returning the report to you
How to try it
Editorial, not from the source
- Name the event and the exact date range to cover, such as specific days of a livestream.
- Point it at the transcripts or notes you already have, if any, so it's not starting from nothing.
- If it needs more sources, connect a research or search tool so it can look beyond what you gave it.
- Review the compiled report yourself, since a research compilation can still misread or miss parts of a transcript.
Give it a defined event and range
The source report describes a specific scope: full deep research on two named days of a livestream event, not an open-ended request to 'look into' it. That specificity is worth copying directly into your own request. Name the event, the exact days or sessions you want covered, and what kind of material counts, transcripts, community notes, or anything else you consider relevant. An open-ended research request without those boundaries risks either missing sessions you cared about or pulling in material from outside the event entirely. Setting the scope up front also gives you something concrete to check the finished report against: did it actually cover both days, and did it stick to the event itself rather than wandering into tangential material?
Decide how much sourcing it needs to do itself
The source account describes wiring up a research API specifically so the agent could pull in more sources when it needed them, beyond whatever transcripts or notes were already on hand. That's a meaningful setup choice: without a connected research tool, the agent can likely only work from what you directly provide, which limits the report to your own material. With one connected, it can look further afield, which broadens the research but also means you're trusting a wider, less controlled set of sources. Decide which situation fits your task. If you already have solid transcripts and notes, pointing it at those directly may be enough; if coverage is thin, connecting a research tool, as described in the source, fills the gap.
Review the report against what you know
The source post describes the job being given and promises a followup thread showing what came back, but the text here stops short of showing the actual report, its length, or its accuracy. That gap matters: there's no established track record here for how complete or reliable a compiled research report turns out to be, only that one was requested and apparently delivered. When your own report comes back, check it against whatever you personally know about the event, or against the original transcripts for a few spot checks, rather than treating a long, well-organized document as proof of accuracy on its own. A compiled report is a starting point for understanding the event, not a verified record of it.
Open Muse to try it. That is the product's own site; this guide is not part of it.
Limits and confirmation points
Limits to confirm
- The source describes giving the job and promises a followup on what came back, but doesn't include the report's content, length, or accuracy in the post itself.
- Nothing here states how long the research took or how it was verified against the original transcripts.
- A connected search tool changes what the agent can find beyond what you give it directly; the two setups are different in scope.
Stop and check
Editorial, not from the source
- Before it starts: confirm the event name and date range.
- If sourcing is thin: connect a research tool rather than letting it guess.
- After it's done: review the report against what you know happened.
If it goes wrong
Editorial, not from the source
- The report misses parts of the event
- Point it directly at any transcripts or notes you have for the missing parts; the source report doesn't describe how complete the compiled result was.
- You're not sure where facts in the report came from
- Ask it to cite which transcript, note, or search result each claim is drawn from.
- It needs sources you didn't provide
- Connect a research or search tool, as described in the source report, rather than leaving it to guess.
Common questions
How long does deep research on an event like this take?
The source report doesn't say. It describes the job being given and promises a followup on the result, but not the time it took to compile.
Do I need a research API connected?
Not necessarily. The source describes connecting one specifically so the agent could pull in more sources when needed; if you already have full transcripts and notes, that may be enough on its own.
Will the report cover every session of the event?
Only if you ask for the full date range explicitly. The source report names specific days; an open-ended request risks missing sessions.
How accurate is the compiled report?
Not established by this source. Review it against transcripts or your own knowledge of the event rather than treating a thorough-looking report as automatically accurate.
Can it find sources beyond what I give it?
Only with a connected research or search tool, as described in the source report. Without one, expect it to work from whatever transcripts or notes you directly provide.
Next
Related to this task
More in Learning
Get recurring task recommendations — ask your own agent to check for relevant updates.