NEWSensighty v2.0 is live — universal subject-as-data engine

Measure real mastery.
Prove it.

Sensighty is universal adaptive learning and psychometric assessment. Generative AI builds the curriculum and questions for any subject — a deterministic Bayesian IRT engine scores every answer, so results are fair, repeatable and verifiable.

Start a free assessment
Any subject, no templates AI never scores SHA-256 verified credentials
Any subjectSubject-as-data, zero hard-coded rules
3PL IRTBayesian ability estimation
AdaptiveEach question matched to the learner
SHA-256Tamper-evident public credentials
Watch on YouTube
The dual-engine core

One platform. Two engines.

Most "AI assessment" tools let a language model grade answers — and grades drift. Sensighty separates the two jobs: AI creates the learning, mathematics measures it.

Generative AI · synthesis layer

Creates the learning

  • Builds curricula and concept graphs for any subject on demand
  • Writes calibrated questions and Socratic micro-lessons
  • Adapts explanations and analogies to each learner
Decoupled
Deterministic kernel · mathematical core

Scores the learning

  • Bayesian three-parameter (3PL) Item Response Theory
  • Repeatable scoring — no prompt drift, no LLM variance
  • Estimates true ability (θ) with a bounded standard error

AI never scores. Mathematics does.

Core capabilities

Built to be trusted

Four design decisions that make Sensighty's results something you can defend to a learner, an employer or an auditor.

Zero-subject constraint

A universal ontology parser turns raw knowledge into a concept graph — no hard-coded templates or subject limits.

Physics, law, Python — same engine

Decoupled Bayesian IRT

AI never scores a response. 3PL IRT calibrates ability (θ), discrimination (a), difficulty (b) and guessing (c).

Zero prompt drift

Socratic remediation

Pinpoints the exact concept a learner is missing and generates a targeted micro-lesson — then retests with fresh items.

No recycled questions

Cryptographic proof

SHA-256 hashed audit ledgers and public credential URLs make every mastery claim tamper-evident and checkable.

Verifiable by anyone
Live demo · try it now

Test-drive the adaptive engine

Answer as a candidate would. Watch ability (θ) and measurement confidence update after every response — exactly as they do in a real Sensighty session.

Latent trait ability (θ)Tier 2 · Intermediate
+0.00standard error ±0.75
−3.0 Novice0.0 Population mean+3.0 Master
Std error (σ)±0.75Calibrating…
Next item (b)+0.00targeted to θ
Questions0continuous update

Simulate the candidate's next answer

3PL logistic IRTdeterministic core
P(θ) = c + (1 − c) / [1 + e−a(θ − b)]
a · discrim1.45
b · difficulty+0.00
c · guessing0.20
Session log 0 events
Pick an answer on the left to see the Bayesian update.
Closed-loop learning

How the adaptive lifecycle works

A continuous loop that diagnoses proficiency, closes knowledge gaps and proves the result.

01

Curriculum synthesis

Any domain is mapped into a knowledge tree: subject → domains → topics → concepts, with prerequisites.

02

Adaptive examination

Each next question is chosen to match the candidate's current ability estimate — no wasted items.

03

Bayesian diagnostics

Mastery is tracked per concept with confidence weights, so a high total can't hide a weak spot.

04

Targeted retesting

Socratic micro-lessons for weak concepts, then a focused retest with brand-new, unseen items.

Why Sensighty

Decoupled intelligence beats the alternatives

How Sensighty compares with traditional LMS testing and prompt-only "AI quiz" tools.

Capability Traditional LMSFixed quizzes & question banks "AI quiz" wrappersPrompt-only LLM tools SensightyGen-AI synthesis + Bayesian IRT
Scoring integrity
Raw percentage

Blind to difficulty — a lucky guess on a hard item counts like an easy one.
Subjective LLM grading

Prompt drift and non-deterministic grading between runs.
Deterministic 3PL IRT

Ability (θ) calibrated with difficulty, discrimination and guessing.
Curriculum
Manual question banks

Weeks of authoring; static items leak and go stale.
One-shot prompts

Flat question lists with no scale or prerequisite structure.
Universal subject-as-data

Automatic subject → domain → topic → concept hierarchies.
Adaptivity
Fixed linear tests

Everyone answers the same sequence regardless of level.
Random variants

Arbitrary difficulty shifts with no ability estimate.
Real-time adaptive testing

Each item selected at the candidate's ability frontier — shorter, more precise exams.
Remediation
Static answer key

Retests reuse questions — testing memory, not mastery.
Generic summaries

No structured diagnosis of specific misconceptions.
Socratic micro-lessons

Targeted lessons for weak concepts, then non-repeating retests.
Credentials
Static PDFs

Easy to edit; no link to actual performance.
None

No credentialing or audit trail.
SHA-256 verified

Public verification URL, LinkedIn export and an immutable audit log.
Who it's for

One engine for every stakeholder

Purpose-built workflows for the people who learn, teach and govern.

Learners & candidates

Personalised adaptive growth

Tests that meet you at your level — no redundant questions, clear next steps.

  • Adaptive difficulty that adjusts to every answer
  • Instant micro-lessons with audio for exact misconceptions
  • Verified badges awarded on proven mastery
  • Fresh retests — never the same question twice
Educators & evaluators

Author a subject in minutes

Define curriculum hierarchies fast and keep full control over item quality.

  • One-click taxonomy of domains, topics and objectives
  • 7-step question verification for keys, ambiguity and difficulty
  • Human-in-the-loop review before anything reaches learners
  • Concept-level analytics that surface hidden gaps
Admins & enterprises

Governance & security

Operational control, audit-grade logging and fine-grained access out of the box.

  • Tamper-evident audit ledger for every security event
  • Role-based access — 6 roles, 22 permissions
  • Session control to revoke devices remotely
  • Live operations dashboard for engine health
Verified credentials

Certificates anyone can verify

When a learner meets the certification standard, Sensighty issues a permanent public certificate linked to the actual assessment record — not a PDF anyone can edit.

Add to LinkedIn

One-click export with certification details pre-filled.

SHA-256 proof

Deterministic hash plus QR scan-to-verify.

Print-ready PDF

Clean landscape certificate for physical copies.

Six-tier badges

Awarded from verified records — never duplicated.

SensightyVERIFIED CERTIFICATE

Certificate of Mastery

Data Structures

Ability θ+1.42
TierMastery
Issued2026
sha256 9f3c2a7be41d08c65f7a91e2d4b3c0a8f16e5d27b9c84a03e1f6d25c7b8a9e40
A product of NEXMANCER● Ledger validated
VERIFIED
✓
Subject catalog

Start with a ready-made subject

Pre-built curricula, ready for adaptive assessment today — or bring your own.

30 questionspy-practical

Python Practical

Hands-on implementation across four real-world domains — algorithms, comprehensions, concurrency and OOP.

1 domain · 6 topics
47 questionscaaip-core

Certified Agentic AI Professional (CAAIP)

Agent architecture and design patterns, autonomous reasoning, tool grounding and multi-agent orchestration.

1 domain · 8 topics
35 questionslib-sci-01

Library Science Basics

Information organisation, metadata schemas, classification and cataloguing fundamentals.

1 domain · 5 topics
Enterprise-ready

Architectural guarantees

Built from first principles for reliability, auditability and psychometric fidelity.

Subject-as-data

No subject-specific code anywhere. Physics, programming, law or medicine run through the same engine.

Multilevel Bayesian mastery

Mastery per concept, topic and subject with confidence — one lucky guess can't make an expert.

Targeted remediation

High totals can't hide localised misconceptions; failed concepts trigger lessons and retests.

7-step quality pipeline

Schema, answer-key, code-sandbox, ambiguity, duplicate and calibration checks on every item.

Configurable psychometrics

Tune questions per round, pass thresholds, remediation triggers and AI provider — globally or per subject.

Deterministic gamification

Badges, streaks and skill radars backed by verified records, not arbitrary points.

FAQ

Questions, answered

Does AI grade my answers?

No. Generative AI is used to create curricula, questions and explanations. Every response is scored by a deterministic Bayesian Item Response Theory (3PL) engine, so the same answers always produce the same result.

What is Item Response Theory, in plain words?

IRT scores a learner by taking into account how hard each question is, how well it separates strong from weak learners, and how likely a correct guess is. The result is an ability estimate (θ) with a stated margin of error — far more precise than a raw percentage.

Can I use Sensighty for my own subject?

Yes. Sensighty treats subjects as data. Describe or upload a subject and it builds the domain → topic → concept structure and a question bank, which educators can review and approve before use.

How are certificates verified?

Each certificate is tied to its assessment record by a SHA-256 hash and has a public verification URL and QR code. Anyone — an employer, a university — can confirm it's genuine and unaltered.

Will learners see the same questions on a retest?

No. Retests draw new, unseen items focused on the concepts the learner struggled with, so they measure understanding rather than memory.

Launch your first adaptive assessment today

Try a ready-made subject, create your own in minutes, or talk to us about rolling Sensighty out across your institution.

Start free Book a demo
Sensighty in 51 seconds

Subject catalog

Ready-made curricula · subject-as-data

Certificate verifier

Check a SHA-256 hash or certificate ID against the ledger

Demo verifier — the sample hash above resolves to a sample record.

Candidate dashboard

Concept-level mastery · sample candidate

Sample Candidate

Certified Agentic AI Professional

θ +1.42 · Tier 3
Core agent architecture94% · θ +1.85
Autonomous reasoning & planning82% · θ +1.20
Tool grounding & sandbox safety61% · lesson scheduled
Retest queued: a 3-question adaptive micro-remediation for "Tool grounding & sandbox safety", using only unseen items.

Illustrative sample data.

Admin console

Governance · access control · audit

Roles6
Permissions22
Audit ledgerSealed
[SEC-001] Ledger block validated · sha256 ok
[AUTH-04] Session CAND-8942 issued IRT token θ=1.42
[RBAC-22] Permission matrix enforced on all endpoints
[SESS-07] Remote device session revoked by admin

Illustrative sample data.