Diri can be used through the AI assistant your organisation already uses: Claude, ChatGPT or Microsoft Copilot. You talk to your own assistant as usual, and it reads from Diri and prepares changes for you. Diri holds the structure, the registers, the approvals and the reports; your AI does the gathering, drafting and bookkeeping.
Bring your own AI requires an agreement with Cyrigo. Cyrigo activates it for your organisation. Contact support@cyrigo.com to get started. See Getting started.
- Run a risk assessment as a conversation. The AI interviews you about a system or process and builds a bow-tie risk assessment with causes, consequences and treatments, ready for you to review.
- Bring in assessments you already have. Give the AI a risk assessment from a spreadsheet, a document or another tool, and it converts it into Diri's structure.
- Follow up treatments. Ask what is overdue for a person or a team, and let the AI update statuses and dates from what you tell it.
- Close framework gaps. The AI reads your active frameworks, finds controls without evidence and proposes evidence from what is already registered in Diri.
- Keep registers in step with other systems. Give it an export from your architecture tool, CMDB or HR system, and it prepares updates to IT systems, vendors and other registers.
- Clean up your data. The AI looks for duplicates, missing owners and inconsistent values, and proposes fixes or files suggestions for you to decide on.
See What to ask your AI for examples.
Your AI connects to Diri through Diri's MCP server (Model Context Protocol, the standard AI assistants use to work with other systems). You connect once by signing in to Diri from your AI assistant and approving what it may do.
- The AI reads the registers you allowed, as your own lists show them.
- The AI proposes changes as a package: new records, updated fields, a whole risk assessment.
- Diri checks the package against the register's rules and shows what is new, what changes, and what is held back and why.
- A person publishes it, or rejects it with a comment, under Import & AI in Diri. Nothing reaches a register before that. A published package can also be reverted.
See Reviewing what your AI proposes.
¶ Security and control
- The AI acts as you. It signs in with your own Diri account and can never see or do more than you can. It works in the organisation and module you have open in Diri, and never reaches another customer.
- It cannot publish, approve or delete anything. Those always need a person signed in to Diri. Approvals and signatures stay with people.
- You choose what it may read when you connect, and you can disconnect it at any time under Import & AI > Tokens.
- Your administrator decides whether AI access is on, which AI assistants may connect, and whether the AI may change colleagues' records or only your own.
- The AI is your organisation's own, used under your organisation's agreement with its provider. Cyrigo does not supply or run it.
Data protection: what the AI reads from Diri is sent to your AI provider (Anthropic, OpenAI or Microsoft) and handled under your organisation's agreement with that provider. Make sure your organisation allows that data to be used with the assistant, and include it in your data protection assessment.