AI for Consultants: Why Most Firms Are Still in Pilot Mode
What keeps consulting firms stuck at the experimentation stage — and what it takes to move past it.
Nearly every consulting firm has "started exploring AI." Some people use it regularly. The managing partner has mentioned it in a team meeting. There may be a Slack channel or a shared folder where people drop links to interesting tools.
And yet, a year into that exploration, most of those firms are in roughly the same place they were when they started. More individual users, maybe. No actual change in how the firm operates.
That's pilot mode. And it has a specific cause.
What pilot mode actually is
Exploration without a destination
Pilot mode isn't a lack of interest or effort. It's a structural problem: there's no clear definition of what "done with piloting" looks like. Teams keep experimenting because there's no shared picture of what they're trying to build toward.
The goal of a pilot is to learn whether something is worth doing. But most consulting firms already know AI is worth doing. The pilot phase is over. What they actually need is an adoption phase, and the two require completely different approaches.
Piloting is individuals experimenting independently. Adoption is the firm deciding what good AI use looks like and making it standard. Most firms conflate the two - they keep adding more individual experimenters and calling it progress.
Why the transition stalls
Three things that keep firms stuck
No shared standard for what good looks like
If every person on the team has a different mental model of how to prompt - different approaches, different habits, different thresholds for when AI output is good enough - you can't share prompts, you can't build on each other's work, and the firm can't improve as a unit. Individual experimentation stays individual. A shared understanding of prompt structure is the prerequisite for everything else.
The AI use isn't tied to specific deliverables
Generic AI training produces generic AI users. When firms say "we want to use AI more," without tying that to specific deliverable types - client memos, research synthesis, proposal drafts - there's no way to know what success looks like or measure whether it's happening. Adoption happens workflow by workflow, not all at once.
Nobody owns the transition
Moving from pilot to adoption requires someone making decisions: which workflows, which tools, what standard, who trains whom. That work doesn't happen on its own. In most firms, it's nobody's explicit job - it falls between the managing partner (who endorses but doesn't operate) and the individual power users (who optimize for their own work, not the firm's).
What the move looks like
From individual experimentation to firm capability
Moving out of pilot mode doesn't require a major technology project. It requires three things:
First, a shared baseline. Every person on the team needs the same mental model of how to write a good prompt - not their own approach, but a firm standard. That baseline is what makes it possible to share prompts, train new hires, and improve over time. Without it, every new person starts from scratch.
Second, a prompt library tied to actual workflows. Not a folder of experiments. A maintained set of prompts for the specific deliverables your firm produces regularly, organized so people can find and trust them. What that library actually looks like is a separate problem from building the individual skill.
Third, someone who owns it. Even one person with explicit responsibility for maintaining the library and keeping the standard current changes the dynamic completely. Without that, the firm keeps piloting indefinitely.
The window
The gap that's opening up
A year ago, the meaningful gap was between firms using AI and firms not using it. That gap has largely closed. Today, almost every firm is using AI in some form.
The gap that's opening now is between firms with infrastructure and firms without it. The firms that have moved from pilot to adoption - shared standard, maintained library, consistent workflows - are improving continuously. Firms still in pilot mode are roughly where they were a year ago, just with more individual users. The technology isn't the differentiator anymore. What builds the gap is the infrastructure around it.