The judgment
AI cannot replace.
A platform that develops it.
Judgment Lab is an AI-coached simulation platform that places you in open-ended, realistic scenarios — then scores the quality of your reasoning, not the outcome. It develops the decision-making and systems thinking that hold their value as AI reshapes the rest of work.
Currently validating with operations and supply-chain instructors and early-career professionals. Not yet open to general registration.
This page is written to be realistic rather than promotional. Judgment Lab is at concept stage. The thesis is sound and the timing is strong — but the right next step is a disciplined validation, not a build. We are currently running Phase 1 of that validation: 20–30 testers from the operations and supply-chain training world. We are looking for instructors, L&D leads, and early-career operations professionals who want to be part of that test. If the loop proves itself, the platform earns a build decision. If it doesn't, we will say so.
AI is breaking our proxies
for human capability.
For decades we have inferred someone's capability from proxies: essays, take-home problems, multiple-choice tests, structured interviews. Generative AI has made almost all of these trivially easy to produce or game. The consequence is not only an integrity problem — it is a measurement and development vacuum.
Organisations and educators increasingly cannot tell, and cannot easily build, the thing they most need: a person who can reason, decide, and adapt when the situation is messy and there is no single right answer.
The skills employers most value — analytical thinking, systems thinking, resilience, decision-making under uncertainty — are precisely the ones that conventional courses, tests, and questionnaires measure and develop poorly. They are best revealed and developed through doing: making decisions under pressure and getting feedback on the reasoning. That is exactly what a well-designed simulation delivers and a textbook cannot.
Practise. Be evaluated on reasoning.
See specific gaps. Repeat.
The unit of experience is an open-ended scenario, not a quiz. The decisive design choice: the platform scores the quality of your thinking — not whether the outcome succeeded.
Existing business simulations grade the outcome — did your virtual company's numbers go up. That is gameable and shallow. The capability that has only recently become feasible is having AI evaluate the quality of the reasoning and explain its judgment back to the learner. This is the technical and product heart of Judgment Lab, and the part no current competitor has locked up.
Eight capabilities. Two foundations.
One defensible framework.
The platform rests on two established bodies of work — the cognitive science of executive function (Diamond), and the WEF labour-market evidence on durable, AI-resistant skills. Together they give a defensible, research-grounded set of capabilities to build scenarios and scoring around.
Why this is a moat in the making. Each loop generates structured data on how a person reasons across these eight capabilities — not just whether they got an answer right. Accumulated across many users, that becomes a proprietary dataset on judgment development that is hard to replicate and valuable in its own right.
The space is active, not empty.
The gap is real and specific.
This is an honest read of what exists and where Judgment Lab sits. Three adjacent bands are populated. Understanding exactly where they sit shows why a develop-first, operations-beachhead wedge remains open.
| Players | What they do | Gap they leave | Verdict |
|---|---|---|---|
| Executive assessment Heidrick Immersive, DDI, Korn Ferry, SHL |
AI / assessor-led simulations for leadership selection, promotion, and succession. Employer-owned. Enterprise-priced. | Not developmental, not individual, not early-career. Summative selection only. | Avoid |
| Pre-hire / workforce assessment Pymetrics / Harver, Vervoe, Anthropos |
Game- and task-based screening with AI auto-scoring of candidates at the hiring gate. | Scores traits or task outcomes, not reasoning quality. No coaching or upskilling loop. | Avoid |
| Higher-ed simulations Capsim, Marketplace, Harvard Business Publishing |
Pre-scripted, model-driven business simulations embedded in courses. 200,000+ subscriptions annually (HBP alone). | Outcome-scored, not reasoning-scored. Instructor-sold, curriculum-bound. Not AI-generative. | Established |
| AI tutoring Khanmigo, Carnegie Learning, adaptive platforms |
Personalised Socratic tutoring on academic subject matter. Knowledge-focused and structured. | Does not develop open-ended judgment or executive function. Different job entirely. | Adjacent |
| Operations & supply-chain sims Littlefield, Zensimu / MIT Beer Game |
Fixed-model, multiplayer, concept-specific decision simulations sold into operations courses and L&D programmes. | Outcome-scored, not reasoning-scored. Not AI-native or adaptive. Proven buyers, dated tools. | The opening |
| Judgment Lab | AI-native, adaptive, open-ended scenarios scored on reasoning quality with coaching prescribed — not a gate, not a course, not outcome-scored. | The develop-first judgment gym that scores how you reason and shows you how to improve. | This |
Why operations and systems thinking
is the right entry point.
The same judgment capabilities transfer across industries and roles — but the buyer, the price, the trust bar, and the distribution channel differ completely across segments. Judgment Lab is starting with one vertical, with one buyer: operations and supply-chain training.
The cheapest honest test that could
kill the idea.
The goal of validation is not to prove the idea is exciting — it is to find out whether it is worth building. Each phase has hard pass/kill criteria. We proceed only if they are met.
Express interest in
the pilot programme.
We are looking for a specific group of people for the Phase 1 closed pilot — operations and supply-chain instructors, L&D leads, and early-career operations professionals who want to test whether the loop works. This is not general access. It is a structured validation with honest feedback loops in both directions.
Pilot participants get full access to the 10-scenario seed library, AI coaching on their reasoning, and a direct line to the product team. In return: structured feedback, a 60-minute debrief session, and an honest report on what the data showed.
This is a concept-stage validation. Participation takes 60–90 minutes across 2–3 sessions over two weeks. It is free. We will share what we find — including if the loop doesn't work.