A Framework for Organisational Intelligence
The full PDF guide with detailed walkthroughs, code examples, templates, and implementation checklists.
One-time purchase • Instant PDF download
Think of an AI assistant not as a decision-maker, but as an incredibly knowledgeable librarian. It knows where everything is, can retrieve it instantly, and can explain context—but the professional remains the expert who interprets and acts on that information.
The artifacts that define how work gets done. Policies, contracts, templates, and mandates. These are the living documents that govern daily operations.
The definitions of entities. Database structures, property registries, risk exposure tables. These provide the framework that organises your information.
The accumulated wisdom. Manuals, guides, institutional memory that explains the "why." This is where tacit knowledge becomes explicit.
search_policy).
| Pillar | Tool Category | Purpose |
|---|---|---|
| Operational Assets | Asset Search & Retrieval | Find and read policies, procedures, templates, contracts |
| Structural Schemas | Schema Exploration | Understand data structures, fields, relationships |
| Reference Documentation | Documentation Navigation | Search guides, manuals, and institutional knowledge |
While Vector Search (RAG) is popular, professional contexts often require Keyword Search.
Why? Precision. A municipal officer needs "Bylaw 2024-037", not something "semantically similar." Keyword search is deterministic, transparent, and incredibly fast.
For organisations in the EU, navigating GDPR, the AI Act, and NIS2 is mandatory. The MCP architecture supports this by design.
The AI retrieves only the specific record needed for the query, not the entire database. This aligns perfectly with GDPR Article 5(1)(c).
Personal data can be explicitly excluded from tools. A "Public Assistant" toolset can differ from a "Staff Assistant" toolset.
Every tool call is logged. You know exactly what data the AI accessed, when, and why. This provides the transparency required by the EU AI Act.
This framework is sector-agnostic. Here is how it applies across three very different domains.
The Challenge: Navigating thousands of bylaws, zoning codes, and permit forms.
The Interaction: A citizen asks, "Can I build a garden shed?" The AI uses search_policies to find zoning rules, get_zone_requirements for setbacks, and get_fee_schedule for costs.
Result: An instant, accurate answer citing specific bylaws.
The Challenge: Complex quantitative models, MiFID II/Basel III regulations, and audit trails.
The Interaction: An analyst asks about VaR model stress testing. The AI uses search_models to find the model specs and search_regulations to check Basel III compliance.
Result: A compliant, sourced explanation of risk exposure.
The Challenge: Ticket sales velocity, artist contracts, and venue logistics.
The Interaction: A promoter asks, "What budget do I need for a 1,500-cap techno warehouse party?" The AI queries search_historical_events for comparable data and get_budget_framework for cost structures.
Result: A data-driven budget based on actual historical performance.
Whether you are managing a city, a bank portfolio, or a music festival, the underlying pattern is identical. Professionals need intelligent access to Assets, Schemas, and Documentation.
The MCP-based approach provides this access in a way that is controlled, auditable, and privacy-compliant. It turns your organisation's specific knowledge into an active intelligence.
The full PDF guide with detailed walkthroughs, code examples, templates, and implementation checklists.
One-time purchase • Instant PDF download