Augmenting Your Investigation with AI

Augmenting Your Investigation with AI argues that mature teams get more from investigation than from another round of detection tuning. Consider a 20-person AML team spending 2 hours per alert on 4 alerts per investigator per day: 1,600 alerts a month. Cut investigation time to 1 hour, and that team can clear 1,600 extra alerts a month, hold output with 10 people freed for harder cases, or halve the pressure on the desk. Matching that gain through detection tuning alone would mean moving an 80 percent false positive rate down to 60 percent, a 20-point shift few teams achieve.

The whitepaper sets out why generative and agentic AI, now accessible without a data science build, make investigation the higher-return target in 2026, and walks through how to start: data integration (what you connect and teach the AI), the business case (which processes to automate first), and the build-or-buy decision.

It also introduces Aurora, Consortix's AI methodology for AML investigation and scenario development. On investigation, Aurora aggregates case data, flags relevant patterns, and drafts case narratives and SARs, with a compliance expert signing off on every output. Aurora runs on SAS Viya and Python, and the methodology transfers to a client's own infrastructure. For teams that want a ready-made deployment instead, the paper covers Lucinity's Luci AI plugin, which runs on top of an existing AML system.

Companion reading: Improving Detection with AI, which covers scenario tuning and alert triage for teams earlier in their maturity curve.