METHOD last engine run: 33 agents · 8 routes · 2 killed · 1 call

An engine runs the search. I make the call.

A human consultant cannot explore eight positioning routes and adversarially test every one of them inside ten days. I can, because I spent a year building the system that does the exploration while I do the judging. Here is the shape of it, in full.

01

The five stages

01
interrogate
The engine reads your business the way a diligence team would: your deck, your site, your win/loss data, plus the intake hour with you. The question bank is pooled from the core body of positioning thought, one sharp set per framework, and it asks only what the research could not fill.
02
arbitrate
The frameworks genuinely disagree: differentiate or be distinctive, invent a category or anchor to one, lead with why or with what. The engine locates your business on each of those tensions and activates only the frameworks whose assumptions match your situation. The judgment is situational arbitration, not framework recall.
03
generate 8 routes
Six to eight distinct positioning routes, each built from the governing frameworks and keyed to verified modern analogs: the recent companies whose move matches yours, with sourced outcomes, including the ones that failed. The options you are already weighing enter this slate as labeled bets, tested as hard as mine.
04
try to kill each
An adversarial pass argues against every route: where the proof is missing, where the claim is saturated, where a bigger rival already owns the ground. Verdicts are recorded, not summarized away. The bets that die stay in the deliverable with the full argument that killed them.
05
score, then the call
Survivors get scored on six weighted criteria with risk held separate, so a bold route is never quietly punished for being bold. Then I make the call: a recommendation with three matrix-cited reasons, the conditions under which it holds, the fallback if they fail, and the tripwires that would tell you it is wrong.
02

The matrix stays neutral. The judgment is signed.

LAYER 1 · THE INSTRUMENT

The scoring is transparent and the weights are printed. You see every criterion, every score rationale, every verifier raise ⬆, and every rejected route. Nothing in the matrix is my opinion; everything in it is checkable.

LAYER 2 · THE CALL

The recommendation is mine and it carries my name. Three reasons cited to the evidence, checkable conditions, a stated fallback, and tripwires. If your own option wins, you will see it win. If it dies, you will see exactly why.

In a market where everyone says the same sentence, the position that wins is the one a skeptical buyer can verify.

03 / THE QUESTION YOU ARE ALREADY ASKING

Did AI write my strategy?

An engine I built runs the exploration and the adversarial testing. I judge the output and I make the call, and the call is signed. You are paying for the judgment that three real repositionings bought, applied to a search no human could run alone in the time you have.

Nobody discloses that the Bloomberg terminal wrote the analysis. The engine is the instrument. Your documents are processed through it, which is exactly why I tell you so in the first five minutes of the intake call, and you tell me what cannot leave your walls.

The consultants hiding their tooling are mispositioned. I would know.

04

What a run produces

Every engagement ships at three altitudes, because the buyer has three jobs: decide, understand, and defend the decision upward.

ALTITUDE 1
The deck

The meeting artifact. Sectioned, client-branded, built to walk a leadership team from market to call in one sitting.

ALTITUDE 2
The report

The complete reasoning, ordered: market map, divergences, the full bet slate, the call, the proof plan, the in-market copy.

ALTITUDE 3
The working papers

The audit trail: every verifier verdict, every score rationale, every source, and every rejected route in full. When the board asks how you know, you hand them this.

See the work