Trustworthy systems. By design.
Career direction is consequential. The way Ukandu produces a result should be transparent, auditable, and respectful of the person on the other end of it.
Six lines we hold to ourselves and to you.
Not aspirations. Operating constraints we have built the product around.
Reasoning you can follow.
The matching engine is deterministic. Every score on the result page traces back to specific, named capabilities. You always see why a role matched and what it leans on. No black box, no opaque retraining.
Inform decisions, never replace them.
A result is a starting point, not a verdict. Pair your result with a conversation with someone working in the field. The engine is a tool you direct, not a system that directs you.
Your data is yours.
We do not sell personal data. We do not share with advertisers. You can export, delete, or update your preferences any time. Australian-hosted storage, encrypted in transit and at rest.
Audit for systematic bias.
The capabilities that drive each role come from observable workplace behaviour, not from historical hiring outcomes. The validation set is reviewed for systematic bias, and the methodology survives scrutiny.
16 and over.
Ukandu is built for users 16 and over. Material is written for that audience. Schools and institutions can run cohorts with appropriate consent flows.
Aligned with the Privacy Act 1988.
We align with the Australian Privacy Act 1988 and the AI Ethics Principles published by the Department of Industry, Science and Resources.
The lines we will not cross.
Explicit commitments about what Ukandu is built to do and what it will not be used for.
- Explain every match. Every fit score is traceable to specific, named capabilities. You can read the reasoning behind every role.
- Protect your data. Encryption in transit and at rest. Australian-hosted storage. No third-party advertising trackers.
- Honour deletion requests. Request data export or deletion via privacy@ukandu.com. We respond promptly.
- Stand behind the method. The capability framework is public. The reasoning behind every match is on the result page. Methodology is reviewed against established workplace data.
- Take feedback seriously. Per-role rating signal feeds into ongoing calibration. You shape the engine that helps you.
- Not a hiring decision tool. Ukandu does not score job applicants for employers. It is a self-service career direction tool.
- Not a data broker. We do not sell personal data, capability profiles, or assessment responses to anyone.
- Not a opaque ranking. No closed model that says "trust us" without an explanation behind the score.
- Not a demographic predictor. Capabilities are not stand-ins for race, gender, age, or background.
- Not a identity classifier. We measure what you can do, not who you are.
Engineered scoring. Language-model assist.
Most of Ukandu is deterministic engineering. A small amount of language-model assist generates plain-English explanations, but never produces the score itself.
The match score is deterministic.
Your capability profile meets each role's capability profile. The engine compares them the same way every time, with no model that “decides” your fit. Two identical profiles produce the same ranking. Always. Every score traces back to specific, named capabilities you can read on the result page.
Plain-English explanations.
A language model is used to turn the deterministic score into readable sentences (“why this fits”, “watch out for”, “a day in the life”). It explains the engine's output. It does not produce the output.
Take a result you can actually defend.
Every fit score on the result page is traceable to specific capabilities. Audit the logic. Share it with people you trust. Use it as input, not as verdict.