Advisors

An advisor (persona) is a reusable expert: a role, a decision‑making posture, and a fixed scope of knowledge. It’s the specialist a Decisions playbook delegates steps to, and the maker that drafts the recommendation behind a proposal. Building advisors is an admin task.

Generate with AI, then refine

Describe the advisor in plain language — “a conservative CFO for cash‑flow decisions, uses SWOT, never recommends layoffs” — and the Generate with AI wizard drafts a structured advisor you can refine. Or fill the fields in yourself.

What defines an advisor

• Name & role archetype — e.g. seasoned Chief Financial Officer. The role shapes how it frames answers.
• Posture — how it thinks: priorities, risk appetite, voice/tone, the management frameworks it applies, and red lines (what it will never recommend).
• Knowledge — the libraries (and/or tags) it may search. The advisor can only ever answer from these.
• External tools — optional MCP connections it may call while reasoning.
• Model & reasoning effort — the chat model it uses (or Auto), and how hard it thinks.
• Visibility — Whole organization · a project · Private · Global (operator only). An advisor’s knowledge must be reachable by everyone it’s shared with — a globally‑shared advisor can only use globally‑shared libraries, and so on.

Using an advisor

• In a Decisions playbook, the orchestrator assigns each step to the best‑suited advisor — grounded in your sources, in order.
• In a Decisions proposal, an advisor drafts the recommendation a human then signs off on.

Note

Advisors are versioned. When you change one, plans and proposals record which version answered, so older decisions stay traceable.