Company memory and AI agents: common questions answered
faq
Plain answers on what company memory is, how agents use it, what stays private, and where humans stay in control.
What is company memory?
Company memory is a shared, connected record of what your team knows: files, notes, threads, web pages, and voice notes, linked by topic, client, and project. See Constella for teams.
How is it different from a folder or a wiki?
Folders depend on someone filing things correctly. A knowledge graph links items by meaning and relationships, so related material surfaces together.
Do answers come with sources?
Yes. Reasoning search returns cited answers, so you can open the original quote, PDF, or page and verify it. Instant mode suits quick lookups; Thinking mode suits harder questions.
What are AI agents here?
Agents are written in plain English, each with a role, expertise, and guardrails. They work inside your rules. Read more about Constella AI agents.
Are agents fully autonomous?
No. People set the goals and limits. Approval steps for anything that matters are on the way. See human-in-the-loop approvals.
Who can see my data?
Access follows the permissions you set. Details are on the trust and controls page.
What if I want to switch AI models?
The model is rentable, but your memory is not. You can change models and your knowledge stays with your company. See avoiding AI vendor lock-in.
Does it learn from corrections?
This is coming soon: each correction is meant to improve later answers for the whole team.
Who is it for?
Enterprises needing governed agents, small businesses wanting a partner that remembers, and individuals who want a smarter home for what they save. Watch the FAQ video.