AI Boardroom

The AI Boardroom example is a board-level AI-governance review scenario. It demonstrates how the platform brings structured board materials into one place, with retrieval-augmented agents to summarise, compare, and challenge the documents.

What it models

A simulated quarterly board meeting that produces:

• An agenda.
• A CEO letter setting the strategic context.
• A CFO Q1 financials update.
• A strategic plan update.
• An acquisition proposal.
• A risk-register update.
• A compliance / audit report.
• A resolution pack.

An AI agent ingests all of these as a knowledge base, then answers questions across them — "how does the acquisition affect Q2 financial guidance?", "which risks would the acquisition aggravate?", "what compliance gaps would the new combined entity face?".

Setup

1. Import the AI Boardroom assets into a workspace.
2. Set the documents' clearance to CONFIDENTIAL (only org admins + owners can read).
3. Create an Agent block, attach the knowledge base, and pin it to a board-meeting workflow.

What's interesting about it

• Document clearance demonstrates limited access to sensitive material — workspace members can see the agenda but not the financials.
• Knowledge-base retrieval demonstrates document grounding — every agent answer cites the source documents.
• Cross-document reasoning demonstrates multi-source synthesis under guardrails — the agent can compare and contrast across docs, but the hallucination guardrail limits unsupported claims.

Sample data

The example ships with these documents in public/static/ai-boardroom/:

• 01-agenda.md — board meeting agenda.
• 02-ceo-letter.md — CEO strategic letter.
• 03-cfo-q1-financials.md — Q1 financial update.
• 04-strategic-plan-update.md — strategic plan refresh.
• 05-acquisition-proposal.md — acquisition proposal.
• 06-risk-register-update.md — risk register.
• 07-compliance-audit-report.md — compliance audit.
• 08-resolution-pack.md — board resolutions.
• sample-transcript.txt — sample agent dialogue.

Related

• Knowledge base clearance — the clearance model the example uses.
• Permission model — how clearance maps to roles.
• DLP — guardrails on agent outputs.