ASN Bank
Designing an AI-assisted self-service customer journey
Summary
Architecting and building a multi-agent AI system to transform customer service, enabling effortless self-service while ensuring complex issues reach the right human expert.
We partnered with ASN Bank to design a comprehensive AI-assisted self-service solution. The engagement covered the full target customer journey architecture, from strategic goal definition and multi-agent design to a phased implementation roadmap and measurement framework.
Services
Industry
The Challenge
Increasing the share of customer contacts resolved without human escalation, without sacrificing quality or trust.
ASN Bank's customer service teams were handling a high volume of contacts across voice and chat, many of which could be resolved through intelligent self-service. But there was no unified architecture for how AI agents should interact with customers, hand over to humans, or escalate complex issues.
Problem 01
No unified AI architecture
- ▪High volume of contacts across voice and chat with no intelligent self-service layer
- ▪No shared framework for how AI agents should interact with customers
- ▪Unclear handover and escalation protocols between AI and human agents
- ▪Missing architecture to connect customer intent with the right resolution path
Problem 02
Resolution quality at risk
- ▪Strategic goal to increase self-service resolution without human escalation
- ▪Quality and accuracy standards must be maintained across all channels
- ▪Customer trust cannot be compromised by automation
- ▪No measurement framework to track resolution success and customer satisfaction
The Approach
We designed a full target customer journey architecture built on two principles: effortless self-service and intelligent human support. Six specialized AI agents form the backbone, with a phased roadmap from human oversight to autonomous operation.
01
Customer journey architecture
Designed the end-to-end target journey with AI handling initial contact, smart escalation with full context, seamless human handover, and post-resolution follow-up across all channels.
02
Multi-agent system design
Designed six specialized AI agents: Intent, Authentication, Knowledge, Insight, Action, and Co-pilot. Each with detailed specifications covering purpose, model architecture, input/output definitions, tone guidelines, and interaction protocols.
03
Phased roadmap to autonomy
Created a four-phase implementation roadmap progressing from Human in the Loop to autonomous operation, gradually expanding AI capabilities as trust and confidence grow.
Impact
A complete blueprint for AI-assisted self-service, from customer journey architecture to agent specifications and a phased implementation roadmap.
6 Specialized Agent Specifications
Each agent documented with a full specification covering purpose, model architecture, input/output definitions, tone guidelines, and interaction protocols for responsible, scalable implementation.
4-Phase Implementation Roadmap
A phased plan progressing from Human in the Loop to autonomous operation, gradually expanding AI capabilities with built-in guardrails and escalation paths at every stage.
24/7 Customer Touchpoints
Multiple easy-to-access self-service touchpoints designed for always-on availability, with a measurement framework and defined KPIs for continuous optimization.
“DEUS delivered a comprehensive architecture for our AI-assisted self-service journey. The multi-agent design and phased roadmap gave us the confidence to move forward knowing we have a responsible, scalable blueprint to build on.”
Head of Customer Service, ASN Bank
Methods
A structured design and architecture approach, from customer journey mapping through multi-agent specification to phased delivery planning.
Customer journey architecture
Multi-agent system design
Agent specification
Phased roadmapping
Measurement framework design