We partnered with Viston to accelerate digital onboarding, strengthen AML, and provide an AI copilot to frontline agents—while meeting local data residency.
Business challenge
Onboarding took 5 days on average, high manual KYC workload, rising fraud attempts, and inconsistent policy answers from agents.
Our approach and solution
Viston built a document intelligence pipeline (OCR, entity extraction, cross-checks), liveness/face match, and a graph-based AML risk scorer.
An Agent Copilot with RAG over internal policies and product docs delivered grounded, auditable responses inside our CRM.
AI applications and benefits delivered
ID doc parsing, sanction/PBO screening, transaction anomaly detection, and knowledge-grounded chat for agents.
Benefits: faster onboarding, fewer false positives, shorter calls, improved compliance traceability.
Cost of implementation
USD 135,000 including data residency controls, MLOps, and model validation.
Time to implement
20 weeks to production across retail onboarding and contact center.
Tools and technologies used
Azure OpenAI (UAE North), Azure Form Recognizer, Face API, Neo4j for AML graphs, Databricks, Elasticsearch, Kubernetes, Kong API Gateway, Vault, Power BI.
Quantitative outcomes
Onboarding time reduced 64% (5 days → 1.8 days).
Fraud loss down 29%; AML detection precision +17%.
Agent handle time down 22%; FCR up 13 pts.
Zero critical audit findings in two reviews.
Key performance indicators (KPIs) tracked
Onboarding TAT, KYC/AML alert precision/recall, SAR volume, false positive rate, agent AHT/FCR, policy grounding citation rate.
Pre- and post-implementation metrics
AHT: 7.9m → 6.2m.
False positives: 42% → 29%.
Grounding citation presence: 0% → 100% for Copilot replies.
SAR conversion (to investigation): +21%.
Stakeholder quotes or testimonials
“The copilot’s citations changed the game with auditors. Every recommendation is traceable.” — Head of Retail Banking
Regulatory or compliance considerations
UAE data residency, ISO 27001, PCI DSS network segmentation, FATF guidance alignment, explainability and retention policies.
