✦ AI Scoping — EBIOS Risk Manager

The AI assistant helps speed up the production of EBIOS RM workshops by generating proposals from the context already entered in the project (business values, supporting assets, risk sources, stakeholders).

Available engines

The assistant is configurable on the AI Configuration page (/gestion/ia), with 8 selectable engines: Gemini, Claude, Mistral, Groq, HuggingFace (experimental), OpenRouter (experimental), Ollama (local, experimental), Custom (free URL/model, experimental). Each engine goes through either the native Gemini format, the native Claude format, or the OpenAI chat completions format (mistral/groq/huggingface/openrouter/ollama/custom).

API key

The API key is stored only in the browser (local obfuscation), never synchronised with the project. Access to the configuration page requires AI to be enabled on your account.

What the AI can generate for EBIOS

• Workshop 3 — proposals for attack paths between a risk source and a business value, based on stakeholders already entered
• Workshop 4 — suggestions for operational scenarios (sequence of technical actions) based on a selected strategic scenario, with likelihood rating
• Workshop 5 — suggestions for treatment measures and an action plan based on the identified risks

Usage

1. Open the AI panel from the ✦ button available on the relevant pages (Strategic scenarios, Operational scenarios, Risk treatment)
2. Check/complete the proposed context (fields already filled in the project are picked up automatically)
3. Run the generation — the proposed results must be validated one by one before being added to the table
4. Rework the suggestions like normal rows (they are not locked)

Review is mandatory

AI proposals are a starting point, never a final validation: rating severity/likelihood and assessing methodological relevance remain the analyst's responsibility.