How the investigation works.
TrustLens is designed as a layered investigation system. The goal is to expose useful signals and uncertainty, not manufacture certainty.
1. Normalize the input
The platform identifies the submitted signal type and stores a normalized investigation record for auditability.
2. Run deterministic checks
For supported URLs, TrustLens can inspect basic technical properties such as HTTPS usage, hostname structure and suspicious patterns.
3. Ask the AI layer
The AI receives the relevant evidence and returns structured category, confidence, indicators, summary and actions.
4. Apply risk rules
Technical signals can contribute points through application-controlled rules rather than allowing the model to arbitrarily choose a final score.
5. Present the report
The user sees a readable report with risk level, evidence, indicators and recommended verification actions.
6. Improve the intelligence layer
Future versions can connect recurring patterns, user reports and reviewed sources into a fraud intelligence graph.
What TrustLens can and cannot establish
A suspicious signal is not automatically proof that a person, business or website is fraudulent. A legitimate website can have weak technical signals, and sophisticated scams can look technically normal.
TrustLens therefore uses risk language and evidence labels. Users should independently verify important claims through trusted channels before sending money, credentials, identity documents or other sensitive information.
Why the separation matters
AI models are useful at recognizing language, context and patterns, but a production trust system should not let a model silently turn uncertain observations into a definitive accusation. TrustLens keeps the deterministic scoring layer separate from the explanatory AI layer.