Compliance analysts are meant to be investigating risk and stopping fraud, yet much of their day is consumed by chasing documents and verifying paperwork. When onboarding business customers, analysts manually confirm ownership structures, review screening results and exchange endless emails to fix missing or inconsistent data. Reboarding often means repeating those same checks.
According to Duna, the stakes are considerable. Compliance failures have triggered regulatory action, billions in penalties and lost revenue, with research from Ribbit Capital showing 52% of customers abandon onboarding applications that take longer than 10 minutes. For professionals at banks and FinTechs, errors can even carry personal liability. Meanwhile, PwC reports roughly a third of financial institutions across EMEA expect AML compliance costs to climb by 10-30% over the next two years.
AI is now redrawing this operating model, moving routine work to software so analysts can concentrate on higher-risk cases where human judgment matters most.
From checkbox to judgment
Today’s compliance function spans first-line analysts reviewing onboarding cases, second-line specialists writing rules and setting risk levels, and leaders shaping how the function scales. AI is reshaping each layer, turning work that was 99% checkbox and 1% judgment into the reverse.
Consider an incomplete onboarding application. Traditionally, an analyst emails the customer, waits, reviews new documents and follows up repeatedly, spending hours before the case is even ready for review. A policy-driven platform can instead request and verify missing information automatically, apply company policy to straightforward cases and escalate only the exceptions. With BCG finding false-positive rates can exceed 90%, removing low-risk reviews frees analysts for enhanced due diligence and complex escalations.
A Head of CDD at a European e-commerce platform said, “62% of cases close first-time right with no follow-up. For a Customer Due Diligence (CDD) team, that is a quality signal. It tells us the policy is doing the work upstream, and the team can focus their judgment where it matters.”
Knowledge that compounds
AI also captures case outcomes and reasoning, so rare judgment calls, such as a VAT number discrepancy with no written policy, become shared knowledge rather than problems solved once and forgotten. Experts, in turn, refine the system with every decision, receiving pattern reports via Slack or querying agents about past rulings.
Crucially, growth no longer demands headcount. McKinsey reports banks typically dedicate 10-15% of full-time employees to KYC and AML. One Duna customer cut drop-off rates by 37% and reduced follow-up cases by 53% without adding compliance staff.
A compliance analyst at a European e-commerce platform said, “The main improvement with AI for us has been the automation. We don’t have to do the screening by ourselves anymore, so the reviews are faster and there are fewer manual tasks.”
Human expertise remains the foundation, but the role is shifting from reviewing every case to improving how AI handles the next one.
Copyright © 2026 RegTech Analyst
Copyright © 2018 RegTech Analyst
