Feedback in AI products is mostly telemetry — users vote, then nothing visible changes.
← Back to patterns
People + AI Guidebook
Memory, Context & Boundary
EmergingFeedback Loop
Steer the next answer with the last one
Most AI products collect feedback as silent telemetry. A real feedback loop closes immediately — the user picks a reason ('too long', 'missed the point'), sees the system write a preference, and the next answer demonstrably uses it. Feedback becomes an active control, not a future training signal.
Metaphor
Calibrated Compass
Interaction
Structured Feedback, Visible Memory Write
Use case
Long-term Assistants, Personalisation, Reducing Repeated Mistakes
Failure mode
Silent Feedback
Control mechanisms
Reason Chips · Free-text Note · Pinned Preference · Visible Update
Reference examples
ChatGPT Memory Summary + Sources · Claude Memory Write-Back · Cursor Bugbot Learned Rules
Interactive Example
The live side-by-side is the comparison.
This prototype already contains the contrast between chat and the pattern response. It stands as the primary comparison surface for this page.
A thumbs-down should change the next answer, not next quarter's model.
In most AI products feedback is telemetry: you vote, the signal disappears into a training pipeline, and nothing visible changes in front of you. Users either give up or invent superstitious rituals — re-typing the same correction every session — because they can't tell whether being heard does anything at all.
Feedback Loop closes the loop immediately. A reason chip ('too long', 'missed the point') produces a visible preference write the user can edit, pin, or unpin, and the very next answer demonstrates the change on a comparable prompt. Feedback becomes an active control surface rather than a deferred promise.
Feedback that visibly steers the next response is a control; feedback that vanishes into telemetry is a suggestion box.
DOPE Lens
How this pattern improves AI interaction quality.
Observability
Each feedback entry produces a visible memory write the user can edit
Directability
Pin, edit, or unpin any preference at any time
Predictability
The next answer uses the preference; before/after is comparable on the same prompt
Explainability
Each preference is labelled with the feedback that produced it
Fit
Where it earns its place.
Use it
when
when
Feedback should visibly shape future answers — not just feed silently into telemetry.
Avoid
when
when
The session is short and feedback would not survive long enough to be useful.
In the wild
Where this pattern already ships.
The pattern is shipping but still maturing across products. Entries below show how different teams have approached it; expect variation as the space settles.
Screenshot pending
ChatGPT Memory Summary + Sources
OpenAI
The book icon under a response opens Sources: which memory shaped the answer, an explanation of why it was used, and a ••• correction. The memory summary itself is editable inline — type a change into it or highlight a line to fix it. No reason chips, and the write is retrospective rather than answer-by-answer.
Visit OpenAI ↗
Claude Memory Write-Back
Anthropic
Memory saves as individual Topics listed in Settings → Memory — open one to read, edit or delete it, and the fix applies to every conversation from then on. You can also tell Claude in chat to remember, change or forget something. On by default for Free, Pro and Max; the old Settings → Capabilities path is gone.
Visit Anthropic ↗
Cursor Bugbot Learned Rules
Cursor
Downvotes on Bugbot comments, replies explaining what was wrong, and "@cursor remember [fact]" on a PR become written rules listed by name, content and scoped paths in the Bugbot repository-rules dashboard, editable or deletable there. Candidates only go active once signal accumulates, and auto-disable on negative signal — the write is visible, not immediate.
Visit Cursor ↗
Research Frame
Why this pattern exists, and where it breaks.
Make feedback a structured loop: reason chips, a visible preference write, and the next answer demonstrating the change.
Chat hides whether feedback influenced anything; users either give up or invent rituals to be heard.
Over-eager preference writes can swing the assistant toward instability; users need to edit and unpin.
Common questions
What people ask about this pattern.
When should I use the Feedback Loop pattern?
+
Feedback should visibly shape future answers — not just feed silently into telemetry.
When should I avoid the Feedback Loop pattern?
+
The session is short and feedback would not survive long enough to be useful.
What problem does Feedback Loop solve?
+
Feedback in AI products is mostly telemetry — users vote, then nothing visible changes.
Why is chat the wrong fit for this?
+
Chat hides whether feedback influenced anything; users either give up or invent rituals to be heard.
Grounded in Research
Not just opinion.
The papers, guidelines, and usability studies this pattern rests on.
Google PAIR team
Google Research · 2021 · refreshed 2026-07
Six-chapter first-party guide. Its Explainability + Trust and Feedback + Control chapters map almost one-to-one onto our Trust & Verification and control patterns; its confidence-display guidance ("determine if, when, and how to show confidence") is the softer default our confidence-signals avoidWhen encodes.
Related Patterns
Curated connections, not just neighbors.
Hand-picked relationships — what this pattern pairs with in practice, what it replaces, what it builds on, and what builds on it.
Often paired with
Autonomy Dial
Feedback Loop steers the next answer; Autonomy Dial sets which actions the agent may take without asking.
Alternative to
Constraint Rules Engine
Both shape future runs: Constraint Rules Engine with hard rules you declare, Feedback Loop with preferences from feedback.
Extends
AI Memory
Feedback is the input, memory the artefact it writes; the loop closes when the user sees that write and can edit it.