An AI agent that helps portfolio managers spot risk signals across large loan books, understand what's happening, and decide what to do next. It was shown at Moody's Summit 2026 in San Diego.
Portfolio managers track hundreds or thousands of loans at once. Risk shows up as a pattern across filings, market signals, and borrower behavior. It's easy to miss manually, and often too late to act on by the time it's obvious. The design problem: make an early signal visible, credible, and actionable, without adding another dashboard nobody checks.
The agent surfaces signals a human might miss, explains what's happening and why in plain language, and proposes a next step, always a suggestion and never an action taken automatically. The portfolio manager stays the decision-maker throughout.
Surface accounts with changing risk signals across the whole portfolio.
Explain what changed and why, with a confidence level and sources.
Suggest a next step, backed by a what-if simulator for milder or worse scenarios.
Figma Make got this from concept to a clickable prototype fast enough to put in front of portfolio managers well before a full build. The hardest question wasn't layout. It was calibrating how confidently the agent should speak, which only real risk professionals reacting to draft language could answer. That's what led to the explicit confidence badges: high for direct, quantifiable signals; lower for pattern-based inference. A PM needs to see that distinction at a glance.
The Early Warning System was showcased at Moody's Summit 2026 in San Diego, drawing notable interest from client decision-makers, a strong signal the design was solving a real, felt problem.