Concept stage · Validation in progress

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.


A note on where this is

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.


The Problem

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.

70%
of employers cite analytical thinking as the single most essential core skill — ahead of all technical skills
WEF Future of Jobs 2025 · 14 million workers surveyed
~40%
of the skills required on the job will change by 2030, with skills gaps named the single biggest barrier to transformation
WEF Future of Jobs 2025
170m
new roles projected globally by 2030 alongside 92 million displaced — the largest workforce transition in a generation
WEF Future of Jobs 2025

The Core Loop

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.

1
Generate
An open-ended scenario with no single correct answer
A realistic operational problem with constraints, incomplete information, and competing priorities — a resourcing crisis, a supply disruption, a build-or-buy decision. You are given context and asked to work the problem.
2
Act
Gather information, make trade-offs, commit to a course of action
You work the problem: ask questions, weigh competing priorities, and explain your reasoning as you go. The platform observes how you gather information and what you consider — not just your conclusion.
3
Inject
Conditions change mid-scenario
A new constraint appears. A plan fails. A second variable shifts. The platform observes adaptability — not a one-shot plan under perfect conditions, but how you recover and adjust when the situation moves.
4
Evaluate the reasoning
AI scores how you thought — not whether you got it right
The AI assesses the quality of your thinking: Did you decompose the problem? Did you anticipate second-order effects? Did you resist the obvious-but-wrong first move? Did you recover well from the setback? The outcome is one data point — the reasoning is the signal.
5
Coach and prescribe
Specific feedback, specific next steps
You receive explained feedback on where your reasoning was strong and where it broke down — tied to the specific capability dimensions the scenario targeted. The platform prescribes the next scenario or micro-lesson aimed at your weakest area. The loop repeats.

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.


What it develops

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.

Executive function (Diamond cognitive model)
WEF durable skills — most valued, least AI-replaceable
Executive function
Working memory
Holding multiple interacting constraints and variables in mind simultaneously.
Revealed by: how many moving factors the person can track before their decisions start to degrade.
Executive function
Inhibitory control
Resisting the obvious-but-wrong first response. Impulse management and assumption-testing.
Revealed by: whether they pause to gather information, or leap to the first plausible answer.
Executive function
Cognitive flexibility
Switching strategy when conditions change. Adapting rather than rigidly persisting with an approach that is no longer working.
Revealed by: response to a disruption injected mid-scenario — adapt or persist?
WEF durable skills
Analytical thinking
Structured decomposition of a messy problem into its parts, with clear prioritisation of what matters most.
Revealed by: how they break down the situation and sequence their attention.
WEF durable skills
Systems thinking
Anticipating second-order effects and ripple consequences — not just the immediate fix but what it triggers downstream.
Revealed by: whether they account for downstream consequences, not just the presenting problem.
WEF durable skills
Decision-making under uncertainty
Acting sensibly and constructively when the information is incomplete and there is no clean answer.
Revealed by: the quality of bets made and how they hedge when data is deliberately withheld.
WEF durable skills
Resilience and adaptability
Recovering, re-planning, and continuing to perform after an approach fails or conditions deteriorate.
Revealed by: behaviour after a plan fails — recover constructively, or unravel?
WEF durable skills
Communication and influence
Justifying and explaining decisions clearly to different stakeholders, making a case under pressure.
Revealed by: the clarity and persuasiveness of the rationale given for choices.

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 Landscape

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

Operations first

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.

Proven buyers already exist. Business schools and corporate L&D already pay for operations and supply-chain simulations. We are not creating a category — that is settled. We are offering a better version of something people already buy.
Clear, honest differentiation. The incumbents (Littlefield, the Beer Game) are fixed-model, multiplayer, concept-specific, and outcome-scored. An AI-native version — adaptive, infinite scenarios, and scored on reasoning — is a genuine "why switch."
Easier to build well. Operations has concrete anchors — bullwhip dynamics, capacity planning, scheduling, disruption response — which make scenarios easier to author and AI scoring easier to tie to an explicit rubric.
Underserved segment. Non-elite universities, TVET providers, and mid-market L&D teams are priced out of the US$5,000-per-month incumbent tools. The operations simulation market has a proven need and a gap at the affordable end.
Seed scenario library — 10 operations scenarios
Scenario 1
The Bullwhip Trap
Consumer demand rises modestly, then reverts. Disruption: a stockout scare tempts heavy over-ordering.
Inhibitory controlSystems thinkingDecision under uncertainty
Scenario 2
Capacity Crunch
Your plant is missing lead-time targets. Disruption: a workstation fails mid-week.
Analytical thinkingCognitive flexibilityWorking memory
Scenario 3
The Single-Source Shock
A sole critical supplier halts for an unknown duration. Disruption: a second supplier hints it may follow.
Decision under uncertaintyResilienceCommunication
Scenario 4
The Expediting Spiral
Everything is "urgent." Firefight, or fix the system? Disruption: a major customer threatens to leave.
Inhibitory controlSystems thinkingAnalytical thinking
Scenarios 5–10
The Forecast Fork · The Migrating Bottleneck · The Service-vs-Cost Squeeze · The DC Ramp-Up · The Quality Call · The Peak-Season Gamble
Full library available to Phase 1 testers.
All 8 capabilities covered

Validation roadmap

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.

Phase 0 · Complete
Founder dogfood
4–8 weeks
Prove the loop on a real user — the founder, on real operations decisions. 5–8 hand-authored scenarios. Minimal interface.
Pass criteria
Measurably sharpens real decisions. Founder chooses to keep using it weekly unprompted.
Phase 1 · Active now
Closed pilot
2–3 months
Does the loop work for strangers in the operations vertical? 10–15 scenarios, rubric-based feedback, 20–30 testers from ops/supply-chain courses and early-career.
Pass criteria
Clear share of testers return unprompted. 2–3 instructors or L&D leads say they would pay.
Phase 2 · If Phase 1 passes
Design partner
3–6 months
Will an institution pay and adopt? One paid cohort with an operations/supply-chain course or corporate L&D team. Cohort analytics. Light integration.
Pass criteria
Paid pilot, evidence of capability improvement, willingness to renew.
Phase 3 · Earned
Scale decision
Only if phases 1–2 pass
Productise and broaden. Content pipeline, integrations, broader scenario generation, billing. New verticals beyond operations. Individual consumer layer.
Condition
Retention and willingness-to-pay thresholds from Phases 1–2 are met. Not before.

Phase 1 · Pilot

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.

Who we are looking for
Operations, supply-chain, or logistics instructors (university, TVET, or corporate L&D)
Early-career professionals in operations, supply chain, or management roles (0–5 years)
L&D leads at companies that run operations or supply-chain training programmes
Willing to complete at least 3 scenarios and provide structured written feedback
General public, executive-level selection, or K-12 — not this phase
Looking for a certification, hiring tool, or performance review instrument — not what this is

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.

Register your interest
We will follow up within 3 working days to confirm your place and send access details.

Your details are used only to follow up on your interest in the Judgment Lab pilot. We do not share them. This is not a marketing list.