Research. Build. Measure.
One method, any problem. Take it to the primary research, not the summary of it. Build the model from the best data available. Test it against something independent, and say where it fails. Then ship the tool, not the deck.
For the people who have to live with the decision after the consultant leaves: owners, executives and boards of operations that run on thin margins, 24 hours a day, with no room for a confident guess.
What we do
Every engagement is the same question asked of a different subject: what is actually true here, and how would we know. The subject changes. The standard of proof does not. These are the four we have taken on so far.
Where to put it, and whether it pays.
Demand, coverage and unit economics for any operation that places assets across a geography. Built from public data so the client can check the work, and validated against something we did not build.
How it should run, and whether it does.
Written operating standards with the evidence behind each one and a KPI that proves it, plus the change work that gets them adopted. Culture read in the numbers before anyone names it.
How people learn it, and whether they did.
Purpose-built learning tools for a specific population, skill and gap. Targets from published norms, mastery gated on evidence across days, every error fed back into tomorrow's practice.
What you know, and whether your AI can trust it.
A knowledge base your organization's AI tools read directly, built from your own documents and decisions, with every claim carrying its source. And the people trained to direct the tools rather than take their word for it.
We tested the published site-ranking models that are sold to operators. They do not predict outcomes, p = 0.86. We published that rather than sell one.
The method
01
The technical report, not the summary of it. No name, number, date or rate appears without the file it came from. Public data wherever possible, so a client can check the work.
02
The deliverable is something that runs: a map you can interrogate, a playbook with a number against every standard, an app that adapts. A deck is a summary of work. We hand over the work.
03
A separate method on separate data, not the data that built the model. Our break-even estimate lands within 1.6% of an independently published actuarial study.
04
Out loud, in the deliverable. Every claim is marked verified, judgment, or something only the client can confirm. A finding that kills an idea is worth more than one that flatters it.
How we work
AI is used here for one thing: reading more material than a person can, and turning it into something a decision can rest on. It does not decide anything. Every output is checked by a human before it leaves, and every claim carries a tag saying whether it was verified against a named source, inferred, or recalled. Volume is what the machine contributes. The judgment is not delegated.
Source documents, project records and provenance notes sit in a structured base the analysis is read out of. Pointed at a defined set of real documents instead of the open internet, a model has very little room to invent, because it is required to name the file behind every line. Anything it cannot source does not make it into a deliverable.
Most organizations are handing a public AI tool a question and hoping. We build the base it should be reading instead: your documents, your decisions, your numbers, with the source on every line, inside your own control. Then we train the people who will use it to direct the tool and check its work, which is the part the vendors leave out.
What we have built
None of this was commissioned. Each one started as a question nobody had answered properly, and each one is live and in use. That is a stronger reference than a logo wall, because every result below can be checked.
An interactive siting and coverage tool. 104 bases attributed across seven states, 565 counties scored for demand, payer mix and access, and a break-even model that says plainly where it stops being reliable.
It flagged Colby, Kansas as greenfield before the base there was publicly announced.
A 34-standard operating playbook across eight pillars, from aviation and clinical through dispatch, billing and business development. Every standard carries its evidence and its KPI. Nothing in it needs a capital budget to start.
Written from the operating seat, by someone who ran a 2.4 million dollar service back to profitability.
Daily reading and math practice that watches what a child gets wrong, works on exactly that, and will not advance until the evidence holds across three separate days. Targets read from the published national norms, not quoted from a summary of them.
Started at home. The founder noticed the problem, went to the research, built the tool and measured it. The method, applied to his own son.
The private research base behind everything on this page. Source documents, project notes and provenance records in one structured place, with the original file kept beside every extract and a verification tag on every claim. It is what makes the figures here checkable rather than just quotable.
49,074 transports over five years, by county, obtained through a state public records request. Measured demand, not modelled. Five more states outstanding.
The founder
Leo Mijares reorganized a 2.4 million dollar organization back to profitability with overtime cut 30 percent, ran business development across five locations, and ran a training center of four thousand students a year, in a sector that is 24/7, regulated to the hour and staffed by people who leave when the culture is wrong. US Army combat veteran and service-connected disabled veteran. Pursuing a Doctorate of Business Administration.
Every model on this site was built by the person who will sit in the room when the number is challenged. There is no analyst two levels down.
Contact
First conversation is to find out whether the method fits the problem. If it does not, we will say so and point you somewhere better.