The AI now knows the company.

Update #1 ended with the Imedex team doing real work in the system the morning after the first demo. This installment is about what we delivered next — and why it came before any visible AI feature: a knowledge base.

Why context comes first

If you want AI to be effective in running and growing a business, it has to know that business — the company, its goals, the decisions it has already made. Otherwise every conversation starts from zero, and what comes back is a generic answer in confident language. So the first AI functionality on this journey is deliberately unglamorous: days after the first demo, we loaded the complete Imedex organisational context into their Healtek instance — a structured knowledge base that every AI agent working with their data now builds on.

What was loaded

Thirty documents across fourteen areas, covering the entire operation: company profile and organisation map; sales processes and key accounts; the service model, including BTK safety-check procedures; logistics, from purchase-to-stock to allocation into the Installed Base; finance and WAC; the Installed Base and its data quality; operational playbooks; policies from MDR compliance to data governance; suppliers; the customer lifecycle; technology scouting; the decision log; KPI definitions; and the operating rules for AI agents.

That last area matters as much as the first thirteen. The context doesn't just describe the company — it sets the rules the AI has to play by.

What changes in practice

Answers respect reality. Ask about clients, devices, service or stock, and the answer follows Imedex's processes, speaks Imedex's terminology — Installed Base, LOT/EXP/S/N, WAC — and respects the traceability chain that runs Client → Location → Device → Activity → Cost/Revenue/Compliance.

Certainty is managed. The AI distinguishes between confirmed facts, assumptions, and items explicitly marked "to verify" — so the team always knows what an answer stands on. It's the same discipline update #1 showed live in the demo, now written into the foundation: deterministic, real-data-only, in an industry where traceability is law and the end of every chain is a patient.

The point of it all

Imedex calls itself a Technology Scout: its job is finding the world's best medical technologies and bringing them to clinicians. Every hour the team doesn't spend re-explaining their company to a tool is an hour spent on that. The AI layer exists so the people can focus on the work only people can do — in an endeavour that saves and improves human lives.

A knowledge base is only useful while it stays true. As the weekly cycles produce decisions, they feed back into the context — so the AI's picture of the company ages with the company, not away from it. We've written up the broader principle in "Your AI shouldn't start from zero every day"; the next journey update will report how the team actually uses it.

Talk to us about a similar journey →

Owner: Jan Safka · Healtek field notes, August 2026