Intelligence Service Building
What Makes an Intelligence Service Defensible
A client acts on your intelligence. Three months later, the deal goes bad and someone asks: “Why did we trust this?” If your answer is “the system said so,” you’re finished. If your answer is “here’s the method, here’s the source, here’s why we flagged it,” you keep the client.
Defensibility is the whole game in intelligence services. Not accuracy — accuracy you can’t promise. Defensibility you can.
Three things make a service defensible:
1. Method, not opinion. A defensible report follows the same steps every time, regardless of who runs it. Same checks, same thresholds, same red flags. When you can describe the method in one paragraph, the output stops being a guess and becomes a process. “We screen every counterparty against litigation records, payment history, and director changes” beats “our model scored them low.”
2. Sources you can name. Every claim should trace back to something a client can point to — a court filing, a filed account, a regulatory notice, an aged invoice. If you can’t name the source, don’t make the claim. The fastest way to lose a professional client is one finding that turns out to be invented. They margin your work; they carry the liability. They need to know the floor is solid.
3. Repeatability. Run the same inputs next month and you should get a comparable answer. Intelligence that swings wildly between runs isn’t intelligence, it’s noise with confidence. Repeatability is what lets an intermediary build the service into their own process and stand behind it month after month.
What this rules out: clever-sounding outputs you can’t explain. A risk score with no breakdown. A summary with no citations. A “the AI determined” sentence in front of a client. These feel impressive in a demo and collapse under one good question.
The discipline is boring and that’s the point. Name your method. Name your sources. Make it repeatable. The firms that win the intelligence game aren’t the ones with the smartest model — they’re the ones whose work survives scrutiny.
Build for the day someone challenges it. That day always comes.
Thoughts on building intelligence services, automation, and business in Malaysia.
More notes