Context Assembly
How can agents select fresh context when information expires and approvals take time?
AI that can explain, pause, and stay within bounds.
Applied AI research for real business workflows. We study how systems use evidence, handle uncertainty, and act within clear limits.
Academic partners
How Figena applies advanced research
Six research questions guide how we evaluate AI for everyday operations.
Evidence & Provenance
How can records stay traceable when evidence and business rules change?
Uncertainty & Review
When should an agent pause, ask a question, or request human review?
Decisions Under Constraints
How can delegated work respect shared budgets, time, and resources?
Permissioned Action
How can authority stay valid across delegation and end when access is withdrawn?
Evaluation Over Time
How can we evaluate recovery, partial execution, and meaning across languages?
Why this matters in real workflows
Four principles guide how we evaluate AI in everyday work.
People Under Pressure
Make the next decision clear, even when attention is limited.
Decisions Under Constraints
Keep timing, cost, risk, and obligations visible.
Explicit System Boundaries
Explain what the system can read, recommend, and do.
Review Before Consequence
Keep consequential changes reviewable and attributable.
