AI capabilities
Where AI is available in OVAITY and what it can access.
Status: Available (in-app assistant and agent features; scope limited by permissions)
Where AI appears
- An assistant in the signed-in app
- Missions that run inside projects you can access. This is more than a chat box.
- People chat on a project, from the project header, which is separate from the assistant
What context it can access
The assistant and agent layers operate within OVAITY authorization:
- Content from projects you can access (membership, ownership, or eligible lab inheritance)
- Not arbitrary organization-wide data by virtue of org membership alone
- External collaborators remain restricted to explicit project memberships
Treat AI answers as assistive. Verify scientific conclusions against primary data and labbook/memory records.
What you can ask it to do
Depending on enabled capabilities in your environment:
- Explain project context you already have access to
- Help navigate analyses, datasets, or next steps
- Run missions on projects you can access
Avoid assuming unconstrained write access, silent permission elevation, or complete recall of all historical org content.
Citations, provenance, approvals, audit
- The notebook and memory are the durable places for reviewed scientific claims.
- Provenance for operators is separate from what researchers see in the project.
Data boundaries
- AI features use server-side model configuration; end users do not paste API keys into the product for core chat.
- Retrieval and tools should respect RBAC. If something is missing from an answer, you may lack access or the content may not be indexed yet.
- Connector and worker secrets remain server-side.
Experimental aspects
Advanced assistant capabilities continue to evolve and may change during beta.