- How to structure a prompt so it produces a defensible finding, not a guess: the number, the time frame, the data fields, and what the write-up should contain
- Ready-to-adapt prompts across the typologies your team investigates most: structuring, money mule networks, account takeover, shell companies, sanctions and PEP exposure, and more
- Where a prompt hits its limit, and how to test one against real alerts before you turn a task on in production
.avif)
AI Agent Prompt Catalogue
The AI Agent Prompt Catalogue for AML Investigations
40+ ready-to-adapt prompts for Unit21's Agentic Task Builder, organized by typology, so your team can go from blank page to working task in minutes.
What you'll see
About this catalogue
As AI Agents take on more of the investigative workload, how you write a prompt determines how useful the output is. This catalogue gives every analyst a clear, typology-specific starting point for building tasks in the Agentic Task Builder, covering both Online Search and Data Analysis prompts.
Whether you're building your first task or refining an existing one, it's a companion to How to Write AI Agent Prompts for AML Investigations, showing exactly what a strong prompt looks like, and where to adapt it to your own program's thresholds, lookback windows, and field names.
No items found.