What an AI readiness audit actually finds
Two weeks of looking honestly at where AI already lives in your company. What the audit covers, what it usually turns up, and why the inventory comes before the strategy.
The most common thing our readiness audit finds is that the company already adopted AI. Nobody decided to. It arrived through browser tabs, personal accounts and one enthusiastic team lead, and it has been processing customer data for months. The recent PagerDuty survey put a number on what we see everywhere: two-thirds of office professionals have used AI tools they believed were not permitted.
So the audit starts with an inventory, not a vision statement. Two weeks, three strands.
Where AI already runs. Sanctioned and otherwise: the tools, the data they touch, the accounts they run under. This part is archaeology and confession booth combined. We are not there to scold; unsanctioned use is the best requirements document your company has ever produced, because people only smuggle in tools that solve real problems.
What your systems and data would support. Retrieval over internal documents needs documents somebody can legally retrieve. Copilots need codebases and workflows in a shape assistants can help with. I once watched a company plan an internal chatbot for a wiki nobody had updated since 2022; preparation, it turns out, is most of the game. Art Nikitin's essay AI tools reward preparation makes that argument well, and his delivery track record is what earned the observation.
What the first quarter should hold. The report ranks pilots by prerequisites actually met, names the governance gaps that would embarrass you in a security review, and ends in a 90-day plan your team can run without us.
The deliverable is deliberately unglamorous: an inventory, a gap list, a plan. Enthusiasm is renewable; it will still be there when the prerequisites are.