Prompt Architecture Build for Global Advisory Firm

Created a structured prompt library and shared definitions layer so engagements could be inherited across teams and regions—with consistent, governed AI outputs.

Global advisory firm formalising governance and knowledge continuity

Scroll

Section

The Challenge

Senior partners held decades of client history in their heads. When they retired or moved, that history evaporated. New partners started relationships from scratch, missing critical context.

Every region had their own way of describing engagements. 'Strategic review' meant different things in London and Singapore. This made cross-border collaboration expensive and error-prone.

The firm was piloting AI drafting tools, but without governance. No one knew who was responsible for AI-generated content, how to review it, or when to reject it. Outputs varied wildly between users.

Section

Our Approach

We conducted a semantic audit: mapping how different regions described the same work. From this, we built a canonical offer taxonomy—shared definitions that everyone could use and extend.

The engagement ledger became the single source of truth for ongoing work. Every major decision, client commitment, and strategic shift was captured with context and authority.

We designed a 12-prompt library covering proposal drafting, email refinement, meeting summaries, and research synthesis. Each prompt was tested across 5+ use cases and refined for consistency.

For AI governance, we created explicit review checkpoints. AI could draft, but humans owned. Every AI-assisted output required sign-off from an accountable partner.

“
The prompt library alone was worth the investment. Our London and Singapore offices now produce the same quality of work using the same language. And I know exactly what to check before anything goes to a client.
Partner
Advisory Practice, Global Advisory Firm

Section