A new kind of operator.
- Takes a brief describing the consequential decision being made.
- Models the buyers a decision will affect — segment, behavior, identity.
- Runs the decision through a calibrated synthetic population.
- Reads the calibration — distribution accuracy, outcome accuracy, scope.
- Ships a recommendation with a confidence interval and an honest scope statement.
- Logs the prediction publicly so the calibration index compounds.
- Refuses to ship any work where the recommendation reduces to "we think."
Five days. One discipline.
The first practitioners.
The founding document.
Most consequential decisions inside consumer companies are made on guesses dressed up as evidence.
A pricing decision gets made because the CEO has a "feel" for what the market will bear. A new product gets greenlit because three friends-and-family said they'd buy it. A campaign gets launched because the focus group nodded politely. The numbers come back two quarters later, the launch underperforms, and someone writes a post-mortem about the things they'd have done differently if only they'd known.
The tools we use to know are broken. Surveys ask people what they will do, then optimise for what they say. Focus groups manufacture consensus through social pressure. Demographic personas — Sarah, 34, urban professional, wellness-oriented — are biographical fan fiction, not behavioural models. Even the best traditional research apparatus is structured to validate decisions that have already been made, not to test decisions before they're made.
We've been calling the alternative "synthetic personas" or "AI research." Both miss the point. The point is that there's a new discipline emerging — one that uses calibrated populations of synthetic humans to test consequential decisions before any real budget is committed. We're calling it Decision Engineering.
A Decision Engineer is a new kind of operator. Not a researcher, not a strategist, not a PMM — though they borrow from all three. Their core skill is refusing to ship a consequential decision on a guess. They take a brief, model the buyers it affects, run the decision through the model, read the calibration, and ship a recommendation with a confidence interval and an honest scope statement attached.
What separates them from the people doing this work today is straightforward.
A traditional researcher delivers insight. A Decision Engineer delivers a recommendation with a calibrated probability of being right.
A traditional strategist tells you what they think. A Decision Engineer tells you what the population of your buyers actually responded to in simulation, and what the historical hit rate is on that kind of forecast.
A traditional consultant points at what went wrong after a launch underperforms. A Decision Engineer points at what's likely to go wrong before the launch ships.
The discipline rests on an analogy. Computer Science is an academic field; Software Engineering is its practitioner application. Decision Science is an academic field — operations research, behavioural economics, decision theory — and Decision Engineering is its practitioner application. Software Engineers don't invent algorithms. They use them, with judgement, to ship working software. Decision Engineers don't invent decision theory. They use it, with judgement, to ship better decisions.
What makes this possible now is that the synthetic-population layer finally works. Not perfectly — anyone telling you that is selling something. But well enough that distribution accuracy on benchmark surveys clears the human self-consistency ceiling 85–90% of the time, and the calibration compounds with every engagement. Five years ago this discipline couldn't have existed. Five years from now everyone serious about consumer decisions will operate inside it.
We're at the early end of that arc. Which means we're recruiting.
A belief that running consequential decisions through a calibrated population is better than running them through a conference room.
A willingness to be wrong publicly when the calibration says you were wrong.
The patience to read confidence intervals instead of cherry-picking favourable verbatims.
And — most of all — a refusal to ship work where the recommendation is "we think." The recommendation has to come with a number, and the number has to come with a scope statement that's honest about its limits.
If that sounds like the operator you want to be, we built Simulatte for you. The platform is the infrastructure. The discipline is the work. Both are worth doing.
We're recruiting the first 100 Decision Engineers. The application form is on this page. The cohort is small on purpose.
iqbal@simulatte.io