Overview
How might we design message pinning so people can trust and act on AI output?
When this pattern fits
- Ideal for long conversations, research sessions, and collaborative chats where users need quick access to key information.
When to skip or lighten it
- Ephemeral chats that are deleted after the session.
- Products that already use a dedicated doc or artifact for the same content.
- Threads where everything is equally important (pins lose meaning).
States
State model coming soon
Key UX elements
Key UX elements coming soon
Anti-patterns to avoid
Pins with no unpin or pin limit.
Pinned content that does not jump to the original turn.
Pins invisible on mobile.
Auto-pinning every AI answer until the pin list is useless.
How products use it
| Product | Implementation |
|---|---|
| Discord | Pinned messages channel for important turns. |
| Slack | Pin messages in channels and DMs. |
| ChatGPT | Saved or highlighted turns in long chats. |
| Microsoft Teams | Pinned conversation items for quick return. |
Implementation
Copy this prompt to generate a production-ready implementation in Cursor, Claude Code, Lovable, or any AI coding agent.
Generate a production-ready implementation of the "Message Pinning" AI interface design pattern.
Pattern Definition:Frequently asked questions
What should users pin in AI chats?
Final decisions, accepted plans, constraints, and answers they will reuse—not every assistant reply.
How many pins are enough?
A small cap (for example 5–10) keeps the list scannable. Archive older pins into docs when needed.
How does pinning differ from memory?
Pins are user-curated bookmarks in a thread. Memory is what the model may recall automatically later.
Should pins sync across devices?
Yes for account-based chats. Show a clear empty state when none exist.