Approach
Every engagement moves through the same three phases, in order — because the methodology ladder only works if you know where a workflow sits before you build anything for it.
The Operational Intelligence Assessment (M1) — fixed-scope, $7,500, 3 weeks. We classify one or two real workflows against the taxonomy: legibility, exception mass, verification cost, risk class. The output is a 90-day roadmap, not a sales deck.
EVERY ENGAGEMENT STARTS HERE — NO EXCEPTIONS
The Assessment's findings determine which of the other nine methodologies applies — an architecture review, an evaluation foundry, an oversight redesign, a legibility program. We only scope what the evidence supports; nothing is sold before it's earned.
M2–M10, SCOPED INDIVIDUALLY — SEE THE SOLUTIONS PAGE →Systems ship with decision provenance, oversight sized to consequence, and a named accountable person — not a dashboard nobody watches. Model and prompt changes move through managed change control, not silent drift.
HUMAN OWNERSHIP DOESN'T END AT HANDOFF
How we actually work
No proprietary platform to lock into. We build in your environment, on infrastructure you already own or can inspect — the deliverable is a working system and the documentation to run it without us, not a subscription to a black box.
MODELS
Whatever your compliance posture allows — frontier or open-weight, swapped without rearchitecting, because scoring logic sits below the model, not inside it.
INFRASTRUCTURE
Deployed into your cloud or on-prem environment where required — nothing routes through a third party you haven't vetted.
EVALUATION
Every deterministic classification is unit-tested and versioned; every model step is evaluated before it ships and monitored after.
See it in the corpus
Start where every engagement starts.
The free Diagnostic, or the Assessment directly.