Diagnostic Debt
Nobody calls BAMPT asking for a better intake process.
They call asking for AI. A chatbot for the website, something that can answer questions at nine at night, because that is apparently when people decide to call an estate planning attorney.
That was the ask from a nine-person estate planning firm outside Denver, and it was a reasonable one. Everyone in their world was building an AI concierge. Why not them.
We took the meeting. We did not take the chatbot request at face value.
There Are Two Kinds of Businesses That Come to Us Wanting AI
The first kind has a narrow problem, and AI happens to be the right tool for it. The second kind has a problem it cannot yet name, and AI is the name it has borrowed, because that is the word everyone is using this year.
Most of the firms we work with are the second kind. Not because they are confused. Because the problem they are actually living inside of is harder to see than the solution everyone around them is selling.
We call this diagnostic debt: the gap between the problem a business can describe out loud and the problem it actually has, plus the interest that accrues on that gap every month nobody stops to ask which is which.
The estate planning firm had been carrying diagnostic debt for about three years, since the year their referral volume doubled and almost nothing else about how they worked was adjusted to match.
The Work Starts Before the Recommendation Does
We don't open an engagement with a deck of findings. We open by doing the work itself, because the fastest way to understand a business is to help fix it, not to watch it from a conference room.
So instead of scoping a chatbot, we sat in on intake calls.
The office manager, who had answered nearly every new-client call since 2016, kept a paper checklist taped inside a binder. That checklist lived nowhere else. On the days someone else covered the phones, the questions came out in a different order, and sometimes the question about existing trusts didn't get asked at all. Client information landed in three different places depending on who took the call: a shared inbox, a spreadsheet, and, for one client, a sticky note that eventually got thrown out with the mail.
None of this came up when the firm first described its problem to us. What came up was: we're good at what we do, more people just need to know it, and we keep hearing AI could help.
Why Not Just Add the Chatbot Anyway
Because a chatbot answering questions at nine at night would have fed all of that inconsistency into the pipeline faster, not fixed it.
Picture a clogged pipe. When water moves too slowly, the instinct is to buy a bigger pump. But a bigger pump pointed at a clog doesn't clear the pipe. It just pushes the backup somewhere worse, faster, and now it's arriving with more force.
Intake was the clog. AI was the pump they wanted to buy first.
You Clear the Pipe Before You Buy the Pump
What we built in the first phase of the engagement was not exciting. One intake form. One place for the answers to land. One sequence of six questions asked the same way regardless of who was covering the desk that day.
The interesting part was what changed once it existed. The gap between a first call and a signed engagement letter, which had been drifting toward two weeks, closed to a few days. A new associate could run an intake call in their first week instead of their third month. And for the first time, the partners could actually see how many calls were coming in and where they were dropping off, because the calls were finally landing somewhere anyone could look.
AI Arrived Only Once There Was Something Worth Pointing It At
This is the part of BAMPT's approach that's easy to say and hard to actually do: AI earns its place. It doesn't get invited in to justify itself.
So it arrived small, and it arrived last. A short summary of each intake form, written for the attorney before the first meeting, so nobody opened a call cold. A first draft of the follow-up email, personalized from what the client had already shared, so the office manager was editing instead of writing from nothing.
Neither of those would have worked six months earlier. There was nothing consistent yet for them to summarize.
The Chatbot They Asked For Still Doesn't Exist
The firm hasn't built the nine o'clock concierge bot. They might, eventually. But the problem it was supposed to solve, a slow and inconsistent front door, was already solved by the time anyone got around to asking whether AI should handle the front door.
The binder is retired now. Nobody kept it as a souvenir. It simply stopped being necessary, the way diagnostic debt tends to disappear the moment somebody finally asks the question underneath the one they walked in with.
Where We Tend to Focus
-
Not every process needs it. We start by finding where AI actually earns its place in the work.
-
Before choosing a tool, we look at how the work actually happens now. That's usually where the real opportunity shows up.
-
The tool gets built around how your team already operates, not the other way around.
-
New tools shouldn't mean new habits. AI gets built into the systems your team already uses.
-
A tool nobody uses isn't a tool. This is where adoption actually happens, not just rollout.
-
A lot of what your team knows lives in people's heads. We help turn that into something the work can lean on.
-
Trust comes from clear boundaries, not long policies. We help set the few rules that actually matter.
-
The right tool depends on the job in front of you, not what's trending. We help you pick what fits.