Introduced the AI Agent Builder: an intuitive workflow canvas for designing custom agents that follow your policies. You start with a builder prompt that states the agent’s goal, then wire in data sources (your own or external search), logic prompts that describe decision rules, human-in-the-loop review points for higher-risk cases, and AI routing steps that branch based on outcomes.

Each agent is transparent by default: before execution you see a checklist of planned actions; after execution you get the agent’s reasoning, the data it used, and the policy references it followed. If you disagree, you can override decisions, and the agent incorporates your correction.

Typical applications include monitoring clients’ websites for industry changes, building an onboarding risk agent that evaluates multiple factors and routes edge cases to analysts, drafting SAR narratives and forms for review, or detecting transactions to/from FATF high-risk countries.

Read more here.