Efficiency Boost for Banks: AI Automates KYC Checks with High Accuracy

Incore Bank, a Swiss B2B transaction bank, has achieved up to 99% accuracy using artificial intelligence agents to extract data from customer onboarding documents. This proof-of-concept project, developed in collaboration with Kyndryl and Google Cloud, demonstrates the potential for AI to significantly streamline compliance processes.

The current KYC (Know Your Customer) process is complex, requiring verification of identities, reconciliation across systems, risk factor analysis, and auditable documentation – all tasks typically handled by human compliance teams. This can result in onboarding taking months for new customers.

How the System Works

Incore’s solution utilizes Kyndryl’s Agentic AI Framework with Google’s Gemini models to coordinate multiple specialized agents:

  • Data Extraction Agents: Automatically pull information from onboarding documents
  • Verification Agents: Check data consistency and validity against internal systems
  • Risk Assessment Agents: Identify potential risk factors based on customer profile and transaction history
  • Audit Trail Agents: Create comprehensive records of the entire process, including AI decision-making

This multi-agent approach mirrors how human compliance teams work – with different specialists handling specific aspects of the review. The key benefit is that each agent focuses on one task rather than requiring a single system to manage everything.

Implications for Financial Institutions

While 99% accuracy in data extraction represents a major advance, it’s important to note this doesn’t mean AI makes final KYC decisions with 99% correctness. Rather, it measures the efficiency of information retrieval – one component of a broader compliance process.

This development comes as banks increasingly invest in AI for productivity gains, yet adoption remains limited in critical areas like KYC. Incore Bank’s success suggests that automating evidence-gathering could free up compliance officers to focus on higher-level risk assessment and decision-making while maintaining human oversight.