Cases/01 · AI investments

Erhvervslivets TænketankClient

Erhvervslivets
Tænketank.RAG for Denmark's leading executive think tank.

Independent Danish think tank·2025–2026·Practitioner: Simon

Erhvervslivets Tænketank is the leading Danish think tank for corporate leaders, boards, and executives — an independent organisation with its own board, its own funding, and the support of business partners and foundation actors. Its analysts produce the insights Danish CEOs and boards rely on when they form a view on the questions facing the country's economy.

Challenge

The team's research process was bound by the speed and reach of human reading. Analysts could go deep on a few sources, or wide across many, but rarely both — and never quickly. The team needed a way to harness scaled data so they could go deeper, wider, and earlier when delivering the insights their senior audience expects.

What was done

A retrieval-augmented generation (RAG) system was designed and delivered, tailored to the think tank's research process. The system retrieves relevant material from the team's own document corpus and external sources, then generates analyst-ready summaries with traceable citations. It was built around the analysts' actual workflow — not a generic AI surface dropped on top of a research practice. Adoption was planned in from the first sprint.

Solution architectureDiagram of the RAG solution architecture: Hunter (news aggregation) and Librarian (knowledge repository) feed an Analyst (creative story generation) which is supervised through a User Dashboard and pushed via Delivery Channels and a Marketeer node, all coordinated by an Orchestration Layer.
Five named agents (Hunter, Librarian, Analyst, Marketeer, Orchestration Layer) wired around a single analyst dashboard — source transparency and expert oversight on every output.

Outcome

70%

Research time reduction.

5×

Sources covered per analysis.

Analysts go deeper, wider, and earlier — the way they wanted to work all along.

Practitioner

Simon — AI implementation, Product, Strategy.

Where this fits

This case maps to 'Our AI investments aren't landing'. The engagement combined an Analysis (defining the right problem to solve with RAG) with Implementation (building and embedding the system into the analysts' workflow).

See the situation →See engagement options →