We do not answer for you. You connect your agent to your own document collection, and it drafts questionnaire answers from your sources — on rails we provide. Your data never feeds a shared model.
For teams that prefer turnkey, the managed EquanimGRC agent is available too. Either way, a human verifies before any answer is final.
Two modes
| Mode | Who runs the agent | Best when |
|---|---|---|
| Your agent (recommended) | You — connected to your document collection via MCP | Sensitive documents; everything stays in your tooling. |
| Managed agent | EquanimGRC | Turnkey; minimal setup. |
How it works
- Agent reads your sources. Your agent (or ours) reads your evidence, policies, and prior answers through the MCP connection.
- Drafts an answer. The agent composes a draft answer grounded in your own documents.
- You review. The draft surfaces in your workspace for your team to inspect before anything is submitted.
- Submit via
answer_question. Your agent (or a human) callsanswer_question(scopemcp:write:answers), which records the draft in your tenant. - Human verifies. A reviewer with the appropriate role confirms the answer is accurate and complete before it is marked final. No agent action bypasses this step.
Trust guarantees
- Your documents stay in your tenant.
- With your own agent, they never leave your tooling.
- No shared or cross-tenant model; answers are drafted from your sources alone.
- Least-privilege scope —
mcp:write:answers— grants nothing else. - Human approval is required before any answer is final. No agent action changes that.
Early access
Questionnaire answering is available to early-access customers. Request access and we will reach out when it is your turn.
See also: Tools & scopes for the full answer_questiontool reference, and Agents & evidence for how agents authenticate and operate within your tenant.