Ejiabor Rita
All projects

2026

Oracle - All Seeing Eye

Oracle is a behavioral intelligence engine. It learns from a user’s observed activity across the ecosystem in order to form an Opinion.

  • PyTorch
  • Plotpy
  • Neural sequence encoder

Oracle is Landscape’s behavioral intelligence engine. It learns from a user’s observed activity across the ecosystem—transactions, jobs, services, applications, completed work, reviews, disputes, repayments, and interactions—to build a continuously updated 10-dimensional behavioral profile called an Opinion.

Instead of asking “Does this person have a traditional credit score?”, Oracle asks:

“What does this person’s behavior consistently demonstrate?”

The resulting behavioral signals can then support opportunity matching, trust assessment, financial decisions, and discovery of hidden opportunities.

How it was built

The Oracle was built around two main pipelines

1. Activity → Opinion

Every ecosystem action is logged as a structured Activity with information such as its type, category, metadata, and outcome. The Oracle aggregates these activities and uses them to estimate 10 behavioral primitives: conscientiousness, integrity, risk calibration, temporal orientation, social reciprocity, emotional regulation, adaptability, agency, cooperativeness, and consistency.

Importantly, the Opinion is explainable: its values can be traced back to the activities that contributed to them.

2. Opinion → Decision

The Opinion is not itself a credit score or final decision. Oracle transforms relevant parts of it into behavioral signals for a particular task.

For opportunity matching, Oracle combines hard requirements + explicit fit + behavioral compatibility. For financial products, it provides evidence such as demonstrated reliability or repayment behavior rather than simply saying approve/deny.