Creative · Community report
Clean up a bloated YouTube watch-later list
Muse went through YouTube playlists to learn its owner's interests and suggest cuts to an overgrown YouTube watch-later list, according to one person on X, who called it a quick way for Muse to get to know them. No count of videos reviewed or removed is given, so review its suggestions before anything is deleted.
What you can get
What it is
Reviewed removal list
A reviewed watch-later list with suggested removals and the reasoning for each, left for you to approve.
It came back right when
- Every suggested removal has a stated reason.
- Nothing was removed without your approval.
- The interests it describes are shown as an inference, not stated as fact.
- You can see what remains after the cleanup, not just what was removed.
- Videos you explicitly want to keep were not suggested for removal.
The checks are editorial, not from the source.
- Grant playlist accessStops for your approval
Access to watch-later and other relevant playlists.
- Review and summarize
Reads the list and drafts an interest summary.
- Suggest removals with reasons
Lists candidates for removal and why.
- Approve and removeStops for your approval
You approve, then it removes only what you approved.
Editorial, not from the source
Works when
- Your watch-later list has grown past the point you ever look at it.
- You want a reviewed set of suggestions, not an automatic purge.
- You're curious what patterns it finds in what you've saved.
Does not work when
- You want everything older than a date auto-removed with no review.
- Your watch-later list is small enough to clean up yourself in a few minutes.
- You'd be upset to lose something it suggested removing without a second look.
Clean up a YouTube watch-later list: review template
| Video or group | Suggested reason | Your decision |
|---|---|---|
| [Video/category] | [Old, duplicate, etc.] | [Keep / remove] |
| [Interest summary] | [Inference from saved videos] | [Confirmed / corrected] |
| [Remaining list] | [What's left after cleanup] | [Matches expectation] |
The source case
On X, @ai_ae_fan says they had Muse go through their YouTube playlists to learn their interests and suggest cuts to an overgrown watch-later list, calling it a quick way for it to get to know them, and noting that Google's own Gemini Spark reportedly said it couldn't help with this. One attributed report, checked on 2026-10-09, with no count of videos reviewed, removed, or kept.
A public report by someone outside this library. Not independently reproduced here.
From the source
Source excerpt
I had @muse go through my YouTube playlists, to learn about my interests, and make suggestions to cull down my overgrown "watch-later" list. It was a great way for her to get to know me quickly (and incidentally, something Google's own Gemini Spark said it couldn't help me with).
Quoted from the original posts. Not written by this guide, and not a result produced here.
Suggested prompt
Bring Access to YouTube playlists, including watch-later. Acts Suggests removals from watch-later. Does not delete anything without your approval. 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
- Access to YouTube playlists, including watch-later
Connected
- The YouTube account
Approved by you
- Approval before any video is removed
How to try it
Editorial, not from the source
- Give it access to your playlists, including watch-later, and say roughly how large the list is.
- Ask for suggested removals with a reason for each, too old, duplicate topic, no longer relevant, not a silent cleanup.
- Review the suggested list before anything is removed, since this task is about suggestions, not deletions, unless you say otherwise.
- Decide case by case, or approve categories of suggestions at once, depending on how much you trust the groupings it found.
Treat the interest summary as a draft, not a verdict
The source report frames this task partly as a way for the agent to learn about the person through what they'd saved, and that framing is worth being a little careful with. A watch-later list reflects moments of passing curiosity as much as it reflects who you are, a video saved at 1am during a rabbit hole is not necessarily evidence of a real interest. Read whatever interest summary comes back as a first pass built from limited, specific evidence, your saved videos, and correct it where it's wrong rather than treating it as an accurate read on you. The value here is in the cleanup it enables, not in the profile itself being correct.
Ask for reasons, not just a shorter list
A watch-later list that goes from hundreds of videos to a handful can look like success regardless of whether the right things were kept, which is exactly why the reasoning behind each suggested removal matters more than the count. Ask for a reason attached to every suggestion, too old, duplicate topic, something you've likely already watched elsewhere, rather than a flat list of titles to approve or reject blind. This also gives you something to push back on directly: if a reason doesn't hold up for a specific video, that's a concrete thing to correct, instead of a vague feeling that the cleanup got something wrong. A list of reasons is also easier to skim quickly than hundreds of bare titles with no context attached.
Decide how much batch trust you're comfortable with
Reviewing hundreds of suggested removals one at a time defeats some of the point of asking for help with an overgrown list, so decide upfront how much you're willing to approve in groups rather than individually. Suggestions grouped by category, old sports highlights, duplicate tutorials, things from a channel you no longer follow, let you make one decision that covers many videos at once, which is faster but also means a single bad grouping affects more of your list. Start with smaller batches until you trust how it's grouping things, then widen the batches as the categories prove themselves to be sensible, rather than approving a huge batch on the very first pass.
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
- One attributed report with no count of videos reviewed, suggested for removal, or actually removed.
- 'Learning your interests' from a watch-later list is an inference, not a confirmed fact about you; treat its read on your interests as a starting point, not a conclusion.
- The source notes that Gemini Spark reportedly couldn't help with this task, a comparison claim this directory hasn't verified.
- Old saved videos sometimes represent things you still intend to watch, not just clutter; a suggestion to remove isn't the same as confirmation it's safe to.
Stop and check
Editorial, not from the source
- Before it starts: confirm access and roughly how large the list is.
- Before anything is removed: review the suggested list and reasons.
- After approving: check the remaining list matches what you expected to keep.
If it goes wrong
Editorial, not from the source
- It suggests removing something you actually still want to watch
- Tell it so, and ask for the reasoning pattern that led to that suggestion to be adjusted for similar videos.
- The interest summary feels off
- Treat it as a first-pass read on a playlist, not a profile of you, and correct it directly rather than assuming it's accurate.
- The list is too large to review all at once
- Ask for suggestions grouped by category, topic, age, duplicate, so you can approve or reject whole groups rather than one video at a time.
Common questions
Will it delete videos from my watch-later list automatically?
No, not by default. The task is suggestions with reasons, reviewed and approved by you before anything is removed.
Can it accurately tell me what I'm interested in?
Treat any interest summary as an inference drawn from a limited, specific set of saved videos, not a confirmed fact about you, and correct it where it's off.
How many videos can it review at once?
Not stated in the source. For a very large list, ask for suggestions grouped by category so you can approve or reject in batches.
What if it suggests removing something I still want to watch?
Tell it directly and ask for the reasoning behind that suggestion to be adjusted, so similar videos aren't flagged the same way again.
Is this better than other AI tools at this task?
Unverified here. The source mentions another assistant reportedly couldn't help with this task, but this directory hasn't tested that comparison itself.
Next
Related to this task
More in Creative
Get recurring task recommendations — ask your own agent to check for relevant updates.