AI does the intelligence. People make the call.
Most of what a consulting engagement costs is not judgment. It is research, benchmarking, documentation, and monitoring: the work that had to be done by people because there was no other way to do it. There is now.
01
Intelligence work. AI runs this.
Labor basin mapping and availability modeling. Wage determination research and benchmarking. Regulatory and case research across jurisdictions. Documentation assembly and consistency checking. Continuous monitoring of filings, deadlines, and program conditions. This work is high-volume, rule-bound, and unforgiving of inconsistency, which is exactly what machines are now better at than people. It runs at a scale and cadence no firm could staff, and it costs a fraction of what it did three years ago.
02
Judgment work. Our people do this.
Which structure to use and which to avoid. How to position a case, and when not to bring it. Which relationships matter, and how to hold them. What to concede and what to defend. When the honest answer to a client is no. None of this is a data problem. It is experience, exposure, and accountability, and it does not automate, because the person making the call has to be answerable for it.
03
Delivered as service.
You are not buying a platform, a license, or a seat. There is nothing to implement and no one on your team to train. You are buying the completed work, delivered to a milestone, with a person’s name on it.
What it means for you
Consultancy-grade rigor at a speed and cost a consultancy structured around billable research hours cannot reach. The intelligence layer compresses cycle time and cost. The judgment layer is what makes the output defensible. Neither one works without the other, and the firms selling you only the first one are selling you a tool and calling it a service.
In eighteen months your incumbent will have these models too. What he will not have is the filing calendars, the recruiting relationships in the basins the crafts actually come from, and a record of what an agency did the last time a structure like yours was tested. The intelligence layer is the cheap part. It was always going to be.
Our moat was never going to be the technology. Everyone gets the same models. It is the distribution, the relationships, and the operating record we are building while the rest of the market is still deciding whether this is a product problem.
If your schedule depends on people you do not have yet, start with an assessment.