Incore Bank Tests AI for KYC
Incore Bank's AI proof of concept reached 99% accuracy in extracting customer document data, aiming to reduce KYC processes from months to days.
Incore Bank has tested an artificial intelligence system that reached up to 99% accuracy in extracting data from customer onboarding documents. The Swiss B2B transaction bank built the proof of concept with Kyndryl and Google Cloud to automate parts of the expensive know-your-customer (KYC) compliance process.
According to the bank's announcement, the system showed potential to shrink a KYC process that can take months down to just days. CEO Mark Dambacher stated that at Incore Bank, innovation must go hand in hand with trust, transparency, and strong regulatory governance. The test explores how much work AI agents can do before a human makes the final compliance decision.
AI Agents Divide Up the KYC Work
The proof of concept uses Kyndryl's Agentic AI Framework with Google's Gemini models. Multiple AI agents are coordinated to handle different parts of the KYC workflow. Some agents extract and check customer information from documents. Others gather supporting evidence and identify potential risk factors. A final group generates an explainable risk score and creates an auditable record for compliance staff to review later.
Google Cloud has described a similar multi-agent approach for KYC. In that model, one agent checks a document's consistency and validity. Another cross-references information with outside sources through web searches. The architecture assigns each agent one specific piece of the process instead of having one system attempt everything at once.
Incore Bank's test provides a specific performance metric for this approach. The up to 99% accuracy figure measures how well the system pulled information from documents. It is important to note this does not mean the AI made final KYC decisions correctly 99% of the time. Data extraction is just one input among several that feed into the ultimate human risk decision.
KYC Automation Lags Broader AI Adoption
The Incore Bank test arrives as financial firms are investing heavily in AI but have largely avoided automating KYC. PYMNTS Intelligence found that a majority of financial services and insurance firms plan to increase their AI budgets over the next 12 months. For most of these firms, productivity gains are a top reason for the investment.
Despite this spending, adoption of AI for KYC, know-your-business checks, and identity verification remains low. Only a small percentage of firms currently use AI for these specific tasks. That is the second-lowest adoption rate among the tasks tracked by PYMNTS Intelligence, ahead of only A/B testing.
| KYC & Identity Verification Adoption | Percentage of Firms |
|---|---|
| Plan to adopt/expand in next 12 months | 65% |
| Current users | 20% |
| Longer-term planned adoption | 90% |
| Top area for near-term interest (reconciliation) | 70% |
The data shows this gap is closing quickly. A significant percentage of firms plan to adopt or expand identity verification and KYC automation within the next year. That figure trails only reconciliation automation, which draws interest from 70% of firms. When longer-term plans beyond the next 12 months are included, planned adoption for KYC automation reaches 90%.
Incore Bank's experiment is precisely the type of project this trend is encouraging. The goal is not to replace the compliance officer. Instead, the AI automates the time-consuming evidence-gathering that traditionally consumed most of an officer's day. The system leaves the final decision, and the accountability that accompanies it, firmly in human hands.
