Every company is obeying a scaling law nobody measures. Revenue is the mass, and cost, capacity, cash and complexity each scale with it at their own exponent. Get the exponents wrong and you grow into a weaker company. No system tells you, because software was built to record what happened, not to decide what should.

Allometry measures those exponents from executed work and enforces a decision at the moment a company commits. The first product is a margin floor on quotes for physical operators: the price clears what the work truly costs today, or it does not leave without a human’s name on it. At our first design partner, a Quebec pallet manufacturer, ninety days of closed jobs surfaced approximately $300K in annualized margin leakage patterns. That goes live October 1.

Three things get built, in order. The Pulse, a continuous reading layer over the systems an operator already runs. The floor as a governed Tool. And the Ledger, an append-only record of expected against realized on every commitment. Everything after that, the company Profile, capital against the Ledger, and eventually agent-to-agent work routing, is an extension of those three rather than a rewrite.

Here is the honest part. There is a large codebase and most of it is a year of exploration. The go-to-market plumbing is real and used. The thing the thesis actually needs does not exist yet: no knowledge graph, no coefficient engine, no ledger, no data platform, no governed model workflow. The first job is a disciplined extraction, not a greenfield build and not a rewrite. You get the full state of the code, liabilities included, after a first conversation.

What we are looking for. Someone who has built a system where correctness mattered more than velocity and can say what that cost. Comfortable deleting code. Reads a specification and argues with it rather than implementing it silently. Wants the object model to be theirs; this is not a role that takes tickets. French and English is useful for the Québec operator base, not required.

The process. A conversation about the thesis, not a screen. Then the architecture brief with the honest state of the code. Then a working session on one real decision, the coefficient pooling problem or the extraction plan. Then references and an offer. No take-home; the working session is the signal.

The context, plainly. One founder, pre-revenue, real operating data from one design partner, a specification written down to the calculation, and no team. You would be the first engineer and the one who decides what the platform actually is.

— Taylor GendronFounder, Allometry · Montréal | New York · taylor@allometry.com