Real experts, inside Claude
Connect Sabli once and ask any expert's mind in your Claude chats. Claude picks the right mind, and the answer arrives as a card with the expert's name on it.
How to run it in Claude
About a minute. You sign in with Sabli; no keys to copy.
- 01
Open Sabli in Claude's directory and press Connect.
Open Sabli in Claude - 02
Or add it as a custom connector with this address:
https://sabli.app/api/mcp - 03
Sign in with Sabli and press Allow. Then ask: “Ask a pricing expert's mind how to price our new plan.”
- 04
Then try one of these:
- “Ask Sabli which minds know about pricing.”
- “Ask a product expert's mind to review this onboarding plan.”
- “Use a growth expert's mind to tell me why signups don't turn into paying customers.”
What you can do in Claude
Ask a question
Get the answer that expert would give, in their method and judgment.
Run a workflow
Structured deliverables, like a PRD or a pricing review, the way they'd do it.
Get a critique
Paste a doc, a page or a plan and get it reviewed against their rules of thumb.
Find the right mind
List who's available, their topics and what each answer costs.
Minds you can ask
All minds
Questions
How do answers look?
As a Sabli card inside Claude: the expert's photo and name, the answer, the mind's version and run ID, and a link to their profile. It follows Claude's light or dark theme.
Does it work in Claude Desktop and on mobile?
Yes. Connectors added on claude.ai are available in Claude Desktop and the Claude apps on the same account.
Is the answer from the real expert?
It comes from the expert's AI mind: an AI built with them from their own methods and past decisions, checked and approved by them, and licensed from them. Every answer says so and carries their name, the mind's version and a run ID.
What does it cost?
Answers use credits from your Sabli plan, the same as on the web: 5 free credits when you sign up, then a Day pass or Plus. Signature minds cost 3 credits per answer.
What does Sabli see?
Only what you or your AI tool send to a mind (the question, brief or text to review). It's kept in your run history until you delete it, never sold and never used to train models. Experts never see your questions, only counts.