The instinct, when you start thinking about AI for your business, is to think about it as a new thing. A new tool. A new platform. A teardown and rebuild. But you rarely need to rebuild the house. Usually, you just need to renovate a room.
Most operational businesses have systems doing real work — a CRM, a project management system, a helpdesk, an accounting platform. AI doesn’t render these tools broken or outdated. They’re still the spine of how the business runs.
The question isn’t “what AI tool should I buy?” It’s “where in the systems I already run are humans exercising judgment that doesn’t actually need to be theirs?”
Example. “Andy magic” is my client’s term for an employee’s work that converts basic lead data into rich information in the CRM. Andy examines the online sources available on a lead and finds the pertinent details. That’s a perfect use case for an LLM — turning long, messy text into structured, digestible data. Let Andy review the LLM output (actually, insist on it), but delegate the text trawling to the LLM.
That kind of intervention isn’t a new AI system. It’s an upgrade to a small piece of work inside an existing workflow. No new software. No training initiative. People keep using the tools they already know, and the AI does work in the background within a clearly defined scope.
This is where AI earns its place — in the spaces of your current process where human “magic” resides.