Docs › Writing Studio › Advanced chat: streaming cancel, long-chat compression, AI memory
Advanced chat: streaming cancel, long-chat compression, AI memory
Last updated June 10, 2026 · 4 min read
Subtle behaviours that come up after a while of using the Coach — streaming cancel, long-chat compression, cross-book memory. Here's how they work.
1 · Cancel a stream
AI replies stream (typed one chunk at a time) — you can stop mid-stream.
| How | When |
|---|---|
| Click the red ■ (stop) button in the input box | The AI is writing and you see it's going the wrong way |
| Start typing a new message | Auto-cancels the current stream |
While the AI is streaming, a red square "■" stop button appears in the bottom-right of the chat input box — click it to cancel the current reply.
After cancel:
- What's already produced stays (AI wrote 3 paragraphs, you stopped at 4 → 3 stay)
- No charge for the remaining tokens (cancelled tokens don't count)
2 · Long-chat auto-compression
When a chat gets long, Slima automatically compresses the older messages into a summary. This is decided server-side — there's no fixed message count or token limit you need to watch.
- Early turns → condensed into a "chat summary" card
- The summary preserves decisions, conclusions, context so the conversation can keep going
Trigger notice
When compaction happens, the panel shows a banner titled "Conversation history compacted" with the text "Due to conversation length, history has been compacted (N → M messages) to continue." (N and M are the actual message counts). Click "OK" (or the X) to dismiss it.
Compaction is automatic — there is no manual "summarise" button, and you can't pin individual messages to exempt them from compaction.
3 · Book vs cross-book memory
Default: the Coach's memory is per-book:
- You told the Coach "the protagonist loves the sea" in Book A → Coach in Book B doesn't know
- Privacy / clean context
To carry memory beyond a single book — your preferences and writing style — use account-level AI Memory (below) instead.
AI Memory (account-level)
Even more general: AI Memory crosses chats too — your preferences, style, recurring phrasing. Configured in Account → AI Memory.
See: AI Memory: the AI's impression of you
4 · Rate limits / reasoning timeouts
Occasionally the AI service is busy and you'll see: "AI service is busy, will automatically retry in N seconds…" (N is set by the system).
In these cases:
- Nothing is lost (your prompt remains)
- Slima retries automatically — you don't need to do anything. If repeated retries fail, it shows "Multiple retries failed, please try again later."