Reducing Human-Agent Escalations with AI Self-Service Across Text and Voice
A Dutch provider of energy, internet, television, and mobile services introduced AI self-service across text and voice in Dutch and English. Grounded in the provider's knowledge base and backed by human escalation, the solution led to a client-reported 30% decrease in the percentage of customer conversations escalated to human agents.
Organization (NDA):
One of the largest providers of energy, internet, television, and mobile services in the Netherlands.
Netherlands
The provider needed to reduce routine demand on human agents while giving customers the freedom to use text or voice
and preserving access to personal support whenever a conversation required it.
Across four service lines, routine questions competed for the same support-team capacity as complex and sensitive cases.
Since the assistant would communicate directly with customers, the quality of its responses also had to be measurable.
Business Case at a Glance
- Operating constraint. Repetitive requests consumed capacity that specialists also needed for conversations requiring judgment or personal attention.
- Intervention. DevRain introduced AI self-service across text and voice, grounded in the provider's knowledge base and backed by human escalation.
- Client-reported outcome. The client reported a 30% decrease in the percentage of customer conversations escalated to human agents.
- Business relevance. The escalation percentage measures demand routed to the human support team, rather than chatbot usage alone.
Business Challenge
The provider wanted to reserve more human capacity for complex and sensitive cases without limiting customers to one
channel or placing automation between them and an agent when personal assistance was necessary.
A customer-facing assistant also creates service and brand risk if its answers cannot be checked consistently.
The solution therefore needed measurable response quality alongside a reliable route to human support.
What DevRain Delivered
DevRain built a bilingual AI support assistant for Dutch and English interactions across both text and voice.
Customers can use it to resolve routine requests through self-service, while conversations needing personal assistance
can be escalated to a human agent.
Customer-Facing AI Controls
The solution combines automation with controls designed to protect service quality and customer choice:
- Knowledge grounding. Responses draw on the provider's own knowledge base so they remain aligned with company information.
- Human escalation. Customers can transfer the conversation to a human agent when they need personal assistance.
- Response-quality evaluation. An AI evaluation framework measures response quality and gives the team a consistent basis for deciding what to improve.
Client-Reported Operational Outcome
The client reported a 30% decrease in the percentage of customer conversations escalated to human agents.
This operational indicator tracks the share of support demand reaching the human team through escalation, connecting
the assistant's performance to assisted-support workload instead of chatbot activity alone.
Technologies
.NET
A free, open-source, cross-platform developer platform for building many types of applications.
Azure Foundry
A unified Azure platform for building, deploying, evaluating, and governing AI applications and agents.
Microsoft Agent Framework
A framework for building AI agents and graph-based workflows with model, tool, memory, middleware, and orchestration support.
Facing a similar challenge?
Tell us about your case, and we will show how a comparable solution could work in your organization —
including the business impact you can expect.