Principles
What we believe — and what we'll tell you we haven't proven yet.
A platform is its convictions. Here are the load-bearing ones — each tied to a capability that actually ships, and each carrying the honest caveat about what’s still open. We’d rather show you the unfinished edge than pretend it’s closed.
five convictions · each tied to a shipped capability · each with its honest caveat
The hill we die on
If everyone else’s AI can’t be made transparent, and ours can — we win.
The problem of organizations is producing scale and efficiency while addressing local needs — and doing it in a way that can be , because humans are acutely sensitive to . Effort is voluntary; people withhold it when they can’t see how an outcome was reached, and rumors fill the vacuum. Black-box AI structurally can’t satisfy that constraint. We build so it can. Transparency isn’t a feature — it’s the moat, and the one thing a black-box incumbent can’t copy.
We’re not saying we automatically win the message — that still has to be earned. But that’s the hill we die on, and every principle below is a way of standing on it.
The principles
Five convictions. Each grounded. Each honest about its edge.
- 01
Compensation is not a religion.
Pay has no direct line to firm performance. Its only performance lever is how it shifts the over time: pay competitively where it matters and the people who matter face fewer viable external alternatives, so your high performers and key roles stay. Managing pay as is a false premise — it neither targets the people who matter nor connects, traceably, to performance. Spend comp where it cuts exit risk; let the risk ride elsewhere. You control your exit rate, or your competitors do.
The honest caveat. The mechanism is calculable — pay percentile vs. the market distribution → viable-alternatives count → exit hazard — but the precise per-segment is not yet calibrated. Calibration is the open work (PAT-179); we will not claim a precise exit-rate prediction before it’s earned. - 02
Every number ships with its method visible.
We own the rigor badge. A figure’s trustworthiness lives in its method — the study design, the test choice, the assumption checks, the , the , the citation — and by default that entire chain is invisible behind one confident number. So we put it on the table: a browsable catalog where every study design, statistical test, assumption check, remedy, and sampling method names the conditions it’s right for, the conditions it’s wrong for, and the cited source that grounds it. No black box.
The honest caveat. A cited method is a correct method, not automatically the decisive one for your dataset. The catalog names the right tool; getting the answer right still depends on the design behind your data. We surface the method so you can judge that — rather than asking you to trust the number blind. - 03
Deconstructed for the professional; concierge for the executive.
One capability set, two ways to buy. The toolbox is the self-serve, for the people-analytics professional — small, drop-in pieces you pick up à la carte and run in your own tools. AnyComp (compensation) and Performix (performance + executive insight) are the executive-oriented products that run it for you — the same capabilities, on your data, org-wide. The professional assembles the ingredients; the executive is served the meal.
The honest caveat. Two tiers, one capability set — the concierge products are not a separate, fancier engine; they’re the same shipped primitives composed and run for you. The self-serve tier is the finish line for most buyers; the diagnostic wizard shows you your own value before you ever buy. - 04
Measure why organizations die.
Companies die the way organisms do — not at random, but as the playing-out of accumulated value-waste. measures the engine, not the financial exhaust: throughput ÷ dissipation → a frailty hazard → , read through a panel of protective and acute , with the waste in your key positions called out so you can act on it. You can’t predict the bullet; you can read the immune system.
The honest caveat. The strong claim — that internal metabolic state predicts corporate death after conditioning on age, size, and financials — is falsifiable, and the Tier-1 mortality study that would prove it is the open challenge. Until it lands, the lifespan band is a-priori-calibrated — a transparent forecast, not a verdict. We publish the estimand and the falsification rule for a hostile researcher to run. - 05
See through the data work.
Black-box AI and code fail silently— a clean-looking table that is quietly wrong, inherited and trusted, poisoning the analysis for weeks. Silent failure is the default mode, not the edge case. So we build the data work to be seen through: every step records what it did (rows in / out, join shape, drops with reasons, coverage and concentration) and tests itself, halting loudly with the observed-vs-expected numbers attached. An agent can’t silently default 84% of jobs to one level and have nobody notice. Never a bare point estimate — always the work behind it.
The honest caveat. This is, today, the tested-data-coding discipline underneath the spokes that ingest and code data — not yet a standalone sold product. Pricing and packaging are unearned, so unstated. The engine ships; the product is a roadmap, named honestly.
Convictions you can read. Numbers you can trust. Edges we don’t hide.
Bring your question. We’ll bring the capability that fits, the method behind it, and an honest account of what we’ve proven and what we haven’t.