CertesianPlan a pilot
APPLIED AI · SIMULATION & ASSESSMENT

Make human
performance
visible.

AI simulation and evidence-linked assessment for professional training. Built through OSCE products. Available for scoped technology partnerships.

Explore a partnership See the evidence
THE CERTESIAN APPROACHFIG. 01
A human interaction
Understand.
With evidence.
Simulate
Assess
Review
Conversation → Evidence → InsightTECHNOLOGY CONCEPT
BUILT FROM REAL EDUCATION PRODUCTSFocused on OSCE today.
Designed for transferable capabilities.Meet the ecosystem
NEW / CERTESIAN LABS

A small investigation.
A useful question.

Can you tell what a witness knows from what the evidence proves? Play a five-minute case built around the same information boundaries we study in professional simulation.

SIX EXPERIMENTS · PLAYABLE NOW

Play. Talk. Discover.

Investigate a case, handle a difficult conversation, help Lumi find the light, or take one of three synthetic clinical stations.

Explore Certesian Labs
01 / THE TECHNOLOGY

Practice is the start.
Understanding is the goal.

Three connected capabilities.
One question: what supports the judgement?

01

Simulate the moment.

Role-based conversations with scenario rules, information boundaries and a changing state.

ScenarioRoleInteraction
FOCUS · OSCE EXPERIENCE
02

Find the evidence.

Connect a performance criterion to what the learner actually said, with clear gaps for review.

↗ Criterion linked
IMPLEMENTED · PRODUCT REPORTS
03

Question the result.

Test consistency, examine disagreement and keep human judgement inside the process.

Supported↗ Review needed
MEASURED · EARLY EXPERIMENTS
02 / TWO PRODUCTS, TWO REPORT POLICIES

Show the evidence.
Respect the assessment.

Explore the difference between a simulated verdict and descriptive feedback. Both start with what the session demonstrated.

PRODUCT-INFORMED EXPLAINERAUTHORED DEMO · NOT A REAL ATTEMPT
THE CONVERSATION Select a criterion →
Patient 01

I’m worried about what happens next.

Learner 02

What is worrying you most right now?

Linked evidence
Patient 03

I don’t know what to expect.

Learner 04

That sounds unsettling. Let’s take it one step at a time.

Patient 05

Okay. That would help.

Learner 06

We’ll go through the next steps together.

NMCMATE · SIMULATION POLICY

Simulated verdict: not passed

Demo rule: all essential criteria met, no red flags. Here, checking understanding is marked essential and unmet. This is our example rule, not the NMC’s marking scheme.

WHY THIS ITEM?

An open question invites the patient to explain their concern.

Source · Learner turn 02

Simplified, synthetic examples informed by the two products’ report implementations. The clickable evidence navigation is an explanatory website feature. It does not claim both products already have this interface.

03 / CERTESIAN RESEARCH

Our experiments
changed our minds.

Recorded findings from the systems behind our OSCE products. Including the fixes we chose not to ship.

RESEARCH #001 · KENT TAN

An AI clinical examiner you can audit.

Information boundaries. Evidence errors. Repeatability. The engineering decisions behind the report.

Explore the research
RECORDED EXPERIMENT · 4 PROMPTS × 5 RUNSDownload aggregate data

Better citations.
A less stable verdict.

We changed the marking prompt to stop it citing reference data as evidence. Select a version to see the trade-off we recorded.

RUN 01PASS
RUN 02PASS
RUN 03PASS
RUN 04PASS
RUN 05PASS
0criteria varied across runs
5reported evidence problems

Shipped prompt. The verdict stayed the same, while citation problems remained.

Author-reported aggregate results from Research #001. One motivating transcript; five runs per prompt. Not a clinical-validity or accuracy benchmark. These experiments were not rerun for this website.

Read the method, decisions & limitations

What changed
when we tested it?

Research journal
RESEARCH #002 / 204 CALLS

The time was right.
The explanation was invented.

A witness defended a correct fact with a source the case never supplied.

Read the finding ↗RESEARCH #003 / SHARED CORE

Not demonstrated.
Or not assessed?

A scope error led to a concrete change in our shared assessment contract.

Read the change ↗
04 / TECHNOLOGY PARTNERSHIPS

Your training method.
Our engineering.

For OSCE training providers, nursing and medical schools, and medical education platforms.

01 / DEFINE

Feasibility & assessment design

Find out which parts of your training workflow AI can usefully support.

  • Scenario and rubric review
  • Technical constraints and risks
  • Scoped implementation proposal
OUTPUT · A GO / NO-GO BRIEF
02 / VALIDATE TOGETHER

Simulation & feedback pilot

Turn a small set of your cases into a practice experience your educators can examine.

  • Role and information boundaries
  • Evidence-linked feedback
  • Educator review and findings
OUTPUT · WORKING PILOT + EVALUATION
03 / CONNECT

Platform integration

Scope how proven capabilities fit into your existing learning platform.

  • Application and data contracts
  • Reporting and access requirements
  • Maintenance and handover plan
OUTPUT · SCOPED INTEGRATION
A PRACTICAL FIRST ENGAGEMENT

3–5 scenarios. One learner group.
One agreed question to answer.

Proposed pilot scope; schedule and fees follow review of your materials. Together we measure review effort, feedback corrections and usefulness before discussing expansion.

External engagements are scoped individually. A self-serve SDK/API and turnkey enterprise integration are not currently offered.

05 / FROM PRODUCTS TO SHARED TECHNOLOGY

One direction.
Multiple places to learn.

NMCMATE and NACMATE inform the core engineering. PASSHHA and QuizSprint are part of the education portfolio; use of a common engine across all products is not claimed.

CertesianShared simulation & assessment technologyIN DEVELOPMENT
Existing product

NMCMATE

Professional exam preparation

nmcmate.com
In testing

NACMATE

OSCE · clinical skills

nacmate.com · in testing
06 / THE ROAD AHEAD

Earn the next step.

A roadmap driven by evidence.
Each phase has a condition for moving forward.

PHASE 01 / IMPLEMENTED

Working systems, with explicit boundaries.

NMCMATE already uses scenario controls, criterion-level marking and a code-computed simulated verdict. NACMATE has a separate descriptive report implementation and is in testing.

  • Actor / marking separation
  • Criterion snapshots
  • Product-specific report rules
ADVANCE WHENKeep the implementation and public capability record aligned.
FOUNDER / KENT TAN

Build it. Measure it.
Publish what happened.

My education-product work led to a focused technical question: what makes AI-assisted assessment useful and auditable? Certesian brings together the systems, experiments and technology partnerships that follow from that question.

Read the engineering note
START WITH ONE TRAINING PROBLEM

Bring your expertise.
Let’s scope the technology.

For teams with real scenarios, a training method and an educator who can review the result.

hello@certesian.com

Prefer to describe the problem first? Prepare a short enquiry here, then send it using your email app. No form data is submitted to this website.

Describe the workflow; please omit patient details and confidential learner records.

THE CERTESIAN THESIS

Better practice.
Clearer evidence.
More informed judgement.

Building a specialist AI technology company.
Starting with clinical skills. Growing through proof.

Discuss your training workflow