The AI Skill Gap That's Actually Hurting Professional Services Firms
It's not about using AI at all — it's the gap between individual users and firms with shared systems.
Ask a managing partner whether their firm uses AI, and almost everyone will say yes. A few people on the team use it regularly. Some use it daily. Results are positive. Time is being saved on certain tasks.
Ask a follow-up question - does everyone at the firm use AI the same way, to the same standard, with access to the same tools? - and the answer is almost always no.
That gap, between individual AI use and firm-level AI use, is where firms are actually losing ground to competitors.
The gap that matters
Individual AI skill is a real and valuable thing. A person who is genuinely good at prompting - who understands how to give AI the right context, who knows when to trust the output and when to push back, who can structure a complex task clearly - will consistently outperform someone who is not.
But individual skill does not compound across the firm unless it is systematized. The person who is excellent at AI research synthesis produces great output. When they are on leave, or change roles, or leave the firm, their capability goes with them. The firm did not gain that skill - one person did.
Firms with shared AI infrastructure do not have this problem. A Skill that one person built works for everyone. A data connection that one team set up is available to the whole firm. The institutional knowledge that lives in the system does not depend on any one person's memory.
Three ways the gap shows up in practice
01 — Output quality varies by who runs the task
When AI use is individual, the quality of AI-assisted output depends heavily on who did the prompting. Senior people who have used AI longer tend to get better output. Junior staff get inconsistent results. The firm has no standard, so there is no floor.
02 — Good work does not compound
When someone figures out a great approach to a recurring task - a client memo format that works well, a research prompt that produces reliable synthesis - that knowledge does not accumulate. It lives in one person's workflow. The next person to do the same task reinvents it from scratch.
03 — The most capable people carry the most friction
In firms without shared infrastructure, the people who are best at AI tend to become the people others go to for help. They field questions, review AI-assisted work, and explain prompting strategies repeatedly. Their capability becomes a dependency rather than something the firm owns.
Why basic AI training does not close it
Most firms, when they recognize the gap, reach for training. Send people to a course. Run an internal workshop. Get everyone to a baseline level of AI fluency. This is the right instinct - a shared baseline matters - but it addresses only part of the problem.
A team where everyone can prompt reliably still does not have shared Skills, shared data connections, or documented workflows. Each person has improved individual capability. The gap between individual improvement and firm-level infrastructure remains.
Closing the infrastructure gap requires a different kind of work: deciding which workflows are worth systematizing, building Skills from your best prompts, connecting AI to your firm's actual data, and documenting the workflows in a way that survives turnover. This is not training in the traditional sense. It is building.
What firms that have closed it look like
Firms that have made this shift tend to share a few visible characteristics. Their junior staff produce output that reflects firm standards from early in their tenure, because the standards are encoded in the tools they use. Their best workflows survive personnel changes, because the workflows are documented and not dependent on any one person. When someone on the team figures out a better way to do something, it goes into the library - where everyone benefits.
The compounding works in both directions too. Every Skill that gets added makes the library more useful. Every data connection that gets set up makes future projects start better. The investment has a multiplier.
If your firm has gotten individuals to a solid AI baseline and wants to build the infrastructure on top of it, Apparatus 202 is where to go. If you are still working on the baseline, 101 covers that well.