Who can see your content

Who actually sees what you write? Three scenarios, one matrix: Slima staff, AI service providers, share-link recipients.

Three scenarios

Scenario 1 · Slima staff

Your usage	Can staff see it?
Only use Slima, no AI features	No internal tool shows your text — the admin backend has metadata only (book title, word count, last edit time)
AI Coach / Beta Readers enabled	Staff still can't see the AI op content; the AI provider sees that request
You report a bug and opt to attach content	Slima engineers see the snippet you opted in (for debugging)

Two narrow exceptions sit outside that table — a read-only view of AI memory entries, and production database access. Both are spelled out under the matrix below.

Scenario 2 · AI service providers

When you enable AI Coach, AI chapter analysis, or AI Beta Readers:

• The request context (your chapter + conversation) is sent to the AI provider
• The provider holds that passage briefly for its own abuse detection, then discards it (per enterprise API terms)
• The provider does not use it for training (per the same terms)

See: AI training policy

Scenario 3 · Share-link recipients

When you enable a share link:

• Anyone with the URL can view the content you've chosen to share
• They don't need a Slima account to view
• Revoking the share cuts access immediately

See: Public share link

Three-scenario matrix

	You	Slima staff	AI provider	Third parties
No AI, no share	✅	❌ two exceptions below	❌	❌
AI enabled	✅	❌ two exceptions below	⚠️ per request, discarded after the abuse check	❌
Share link on	✅	❌ two exceptions below	(only if recipient uses AI to view)	✅ anyone with the URL

The two exceptions. No internal tool displays the text of a manuscript, and that is what the ❌ above means. Two narrow cases exist anyway, and they are written here rather than left for you to find:

1. A read-only view of AI memory. When someone reports that the AI invented something that isn't in their book, an engineer can open that book's AI memory entries to trace the fault. Those entries are AI-generated summaries, not your prose, and you can clear them yourself.
2. Production database access. A small number of engineers hold direct access to the production database. It is least-privilege, logged, used only to investigate problems that have been reported, and internal policy forbids using it to read anyone's work.

Neither is a way to browse manuscripts, and neither happens unless something has been reported.

Related

• Where your data lives
• AI training policy
• Slima Privacy Policy