Explainable machine learning · Johannesburg

We tell you whether a market's price is honest.

Aletheia Labs builds machine-learning systems that find what others miss and explain every finding. Our flagship, Sentinel Trace, detects insider trading and manipulation on prediction markets and betting odds.

What we do

Integrity you can show to a regulator.

Every alert carries its evidence. Every decision is recorded. A person always makes the final call.

Flagship product

Sentinel Trace

Market-integrity monitoring for prediction markets and sports betting.

  • Nine detection signals scored together
  • Stays silent when evidence is thin
  • Human review desk and audit trail
Public

Integrity Index

An independent, public score of how clean a market is, starting with South Africa's 2026 local government election markets.

  • Method published before results
  • Built for newsrooms and the public
For companies

Applied machine learning

Explainable models built on your own data, starting with a short paid pilot.

  • Anomaly and fraud detection
  • Risk scoring and early warning
  • Forecasting and prioritisation
Assurance

AI assurance and audit

Independent checks that your models are accurate, fair and still working as intended.

  • Bias and fairness testing
  • Drift and performance monitoring
  • Explainability reports for boards and regulators
  • Governance for AI and agentic systems
Advisory

AI readiness and strategy

A clear view of where AI will pay off in your organisation, and what it takes to get there.

  • Readiness and data assessment
  • Use-case selection and roadmap
  • POPIA-aware data practices
Training

AI literacy and training

Practical sessions that help leaders and teams use AI well and question it wisely.

  • Executive briefings
  • Team workshops
  • Responsible-AI training
Research

Research first, built to ship.

Our products start as research questions. We publish our methods so others can test them.

Market integrity

Detecting trading on advance knowledge in prediction markets and betting, and telling it apart from skilled forecasting.

Model monitoring and explainability

Methods that detect when a model's reasoning changes, not just its accuracy, before decisions go wrong.

Fair and accountable AI

Monitoring how models treat different groups of people, with evidence a regulator can follow.

Publications and reports

South Africa Election Markets 2026: Market Integrity Report
Aletheia Labs · Forthcoming, November 2026
Monitoring Explainability: A System to Detect Changes in Model Feature Importance and Underlying Relationships
Founder's MEng research · Stellenbosch University
Who we serve

Where trust in a number matters.

Prediction markets and bookmakers

Surveillance and an independent second line of review.

Regulators and public sector

Evidence on offshore markets and the AI systems you oversee.

Media and newsrooms

Know whether the odds you quote have been moved.

Financial services

Fraud detection, client risk and model assurance.

Healthcare

Claims, demand and patient-access analytics.

Logistics, property and education

Early-warning and forecasting models on your own data.

How we work

Built to be trusted.

Explainable

Every score shows the reasons behind it.

Human in the loop

Our systems flag and rank. People decide.

Auditable

An append-only record of every finding and decision.

POPIA-aware

Personal data handled under written agreements, with access by role.

Who we are

Aletheia is Greek for truth.

Aletheia Labs is a Johannesburg company building market-integrity tools and applied machine learning for South Africa and beyond.

Sibahle Nela Ngwenya
Founder & CEO
MEng (Data Science), Stellenbosch University · BSc Metallurgical Engineering, Wits
Pilot with usA short paid pilot on your own data.
Research with usUniversities and labs working on integrity and explainable AI.
Invest with usWe are raising our pre-seed round.

Talk to us

Investors, clients, newsrooms and regulators: we'd like to hear from you.