Client privacy in an AI memory: how project separation works

Consultants handle material that clients expect to stay confidential. Before adding it to any AI tool, you need to know how separation and retention work.

Separation by project

Each client lives in its own project. Sources, answers, and context stay within that project, so switching from one account to another does not carry details across.

Retention and training

Under zero data retention agreements, client data isn't stored or used to train the models. The trust and controls page covers the specifics.

Human control

• Drafts wait for your review
• Nothing is sent without you
• You decide what gets captured

A short privacy checklist

1. Check your client contract for rules on third-party tools
2. Keep one project per client
3. Capture only what you need
4. Review the trust page before adding sensitive files

For larger teams

If you work within a firm, see Constella for enterprise for governance options.

Learn more about Constella for consultants or try it at askconstella.com.