I listed some suggestions below. I am open to learn anywhere I am short-sighted --if you read this–and know better, please share!
-
(deleted)
-
In-thread search is broken.
Using Ctrl+F in huge threads is slow and manual. Need scoped, semantic, fast search — across and within chats. -
Default thread labels are useless.
Have first 5 words before return label chat. No way to track evolving content. We need editable labels early in the thread – not later. -
No summaries or outlines.
Chats shift fast. No structure, no memory map. Give us live summaries or outlines, visible when returning to a thread at top or bottom. Important for power users. -
No visible timestamps.
Chat dates are difficult to find. Must show creation + last updated time at top or bottom of thread. -
Model can’t track time.
It guesses wrong even when told the correct date/timezone repeatedly. Allow a time zone setting please!! Breaks trust for time-sensitive users using timed data collection. -
Search is shallow.
Only matches exact words. No idea-based matching. Needs semantic search + date filter + content preview. -
Search results drop dates.
You briefly show a timestamp in results — then it disappears. Dates must be sticky and visible. -
No search & replace.
Can’t redact names or sensitive info across chats. Huge security gap. Needed for compliance, privacy, and publishing. -
**Allow for poor chat response deletion–**for a power user this allow us to keep clean data.
-
Allow for library item deletion – sometimes bad images are generated.Don’t force us to keep. Removal allows for less storage for you, too.
-
Memory is a black box.
No visual cue when memory is in play. Can’t verify what’s remembered or forgotten. Large user can’t find easily. Breaks flow. Can’t verify easily–memory or hallucination. Is this a brief summary in generalized AI form, or accurate quote-- user is in the dark. -
Fakes continuity.
When memory fails, it pretends to know. Just say: “I lost context.” Don’t fake understanding. -
No tagging or filtering.
Manual folders exist but no tag support. No sort, group, or multi-topic organization. Useless at scale. -
No pinning or highlights.
Important insights get buried. Let users pin key messages or mark moments inside threads, into different area, at top of threads. Loss of gold continually. -
No export/snapshot tools.
Copy/paste isn’t enough. Need export by segment, summary snapshots, or downloadable briefs. Can do manually but time consuming. -
No content visibility inside folders.
Folders don’t preview threads. Have to open each one. No glanceable content or summaries per folder. -
Model responses get slower and inaccurate over thread.
Long sessions degrade output quality. No warning, no reset. Shorten length allowed? Alert user? -
No feedback on system behavior changes.
When model behavior shifts, there’s no warning. Did you throttle it? Unthrottle it? Switch personalities? Am I in a test feature now? Removed?? Am I doing something? Not seeing changed features makes users go stir crazy. Perhaps for qualified test users–offer opt in feature–and allow notification on elements where you can. I see shifts frequently and have to figure out what is going on. A few have made me question myself intensely.
Power users are not edge cases.
We’re deep thinkers, fast switchers, and collaborative builders. Let us operate at depth and speed. Give us the structure, access, and signal clarity to stay in flow. I’m loving AI and can’t get it fast enough!! Thank you for all you do!!