The exciting version gets demoed. The boring version gets used in month nine.
There is a whole industry selling the exciting version of this right now. Agents that run your firm. Transformation decks. The demo always works. Gartner has a name for a good share of it: "agent washing", chatbots relabeled as autonomous agents to ride the category.
I build the version that's still working in month nine.
My clients are boutique law firms, CPA firms, nonprofits, and advisory practices. They are not looking to be first. They are looking for something that produces the same output every time, that a supervising partner or director can stand behind, and that will survive a question from opposing counsel, a regulator, or an auditor about how the work was produced.
That is a different product than the one being sold at most conferences. It is less impressive in a demo and considerably more valuable in practice.
What boring means in practice
One process, start to finish. Client intake. Engagement letters. Monthly close prep. Firm-wide initiatives produce announcements. Named workflows produce results.
What client data may be entered, into which tools, by whom. Firms without a policy do not have less exposure. They have the same exposure and no record of a decision.
A prompt that only works when the person who wrote it runs it is not a process. It is a party trick. Repeatability is the whole point.
Someone owns the output before it reaches a client. That person is identified in writing, not assumed.
Hours consumed, error rate, turnaround time. You cannot prove improvement against a number nobody wrote down, and unproven improvement does not survive the renewal conversation.
What was produced, by which process, reviewed by whom. On an ordinary day this is filing. On the day someone asks, it is the entire defense.
None of this photographs well. All of it is why the thing is still running next year.
Why most pilots die
Every one of these is fixable, before the money is spent. Almost nobody checks.
This is not just a hunch. Thomson Reuters titled its own 2026 professional services research "AI's ROI Remains Elusive." That is the trade press of your own industry saying the same thing.
The distinction that matters
The exciting version asks: what can this technology do?
The boring version asks: which of our processes is broken, and can this fix that one?
The first question produces a purchase. The second produces a result. Every failure mode in the research above traces back to a firm that asked the first question.
How the work runs
We pick apart one process that's holding up traffic and establish whether it is a real candidate. Some are not. We will tell you. Stops here if you want.
The workflow gets rebuilt, quickly, with the baseline, policy, prompts, and approval step that make it repeatable by someone other than me. You get a written memo covering what changed. Stops here if you want.
If there's a second problem, a knowledge gap or a policy gap, that's where we go next. Nothing here is a foregone conclusion.
Who this is for
About
I remain a skeptic about most of what is being sold today. A narrow slice of this technology is genuinely transformative inside professional services work. Telling the difference is the job.
That's what Aspen Management Group does: a boutique AI workflow consultancy for small and mid-sized professional services firms.
Exciting is how you end up explaining yourself to a committee.