Automate intent discovery and resolution with Customer Intent Agent

With the latest July 2025 updates to the Customer Intent Agent, Microsoft is accelerating the journey toward fully autonomous contact centers. These enhancements empower organizations to leverage AI agents that not only understand customer intent but also automate the entire resolution process. This increases the availability of human service reps to focus on high-value interactions.

The July preview introduces powerful new capabilities for the Customer Intent Agent, building on the foundation of intent mining, dynamic interviews, and knowledge-driven automation. Now, organizations can achieve more accurate intent detection, seamless handoffs, and flexible integration with custom workflows with the same enterprise-grade control and compliance.

More conversational, capable, and configurable

The Customer Intent Agent’s conversational abilities have been significantly enhanced. Natural language instructions at the organizational and intent level guide the agent’s behavior, enabling more natural, dynamic, and context-aware interactions. The system now supports instructions written in natural language. This allows for dynamic interviews, conditional logic, and seamless transitions between related intents.

Key improvements:

Dynamic interviews: The agent can persist information from self-service, suggest next questions, and handle intent shifts or interruptions gracefully.

Instruction-driven resolution: Playbooks provide step-by-step guidance for resolving intents, leveraging both knowledge articles and connected actions.

Intent-level mapping: The agent can map intents to knowledge, actions (via connectors), or custom AI agents. This enables tailored resolution logic and flexible integration with proprietary systems.

Improved disambiguation: The agent asks clarifying questions and provides immediate feedback. This ensures customer intent is accurately captured, even across thousands of topics.

Voice support

The Customer Intent Agent now supports voice for self-service as well as chat. It uses agentic AI techniques to provide a higher quality of self-service and lower chance of escalation by providing a natural flow of conversation, focusing on narrowing down the customer’s reason for calling and identifying the best solution.

Key improvements:

Natural conversation: The agent speaks in a natural empathetic back-and-forth manner, without robotic prompts. It’s just a conversation that flows, reducing the chance of the call escalating.

No pre-programming: The agent dynamically plans the conversation without pre-programmed flows. It reasons over the customer’s speech, narrowing down the intent and identifying the most appropriate solution.

Robust speech recognition: The customer speaks naturally, as they would to a person, and the agent understands.

Responsive: Fast, natural responses keep the conversation flowing, which is critical in a voice conversation.

Smarter intent discovery and management

A core advancement in this release is the overhaul of intent discovery. The Customer Intent Agent now leverages business context, product hierarchies, and industry profiles to generate more relevant, manageable, and actionable intents. Fewer, higher-quality intents make it easier for admins to approve, merge, and segment them by line of business. The new analytics dashboards and semi-autonomous approval flows help organizations quickly identify and promote the most impactful intents. This streamlines the management process and ensures that intent libraries align with real customer needs.

Key improvements:

Business context-aware mining: Intents are now generated with additional context, reducing noise and improving accuracy.

Granular control: Admins can filter, approve, or merge intents in bulk, and segment libraries by line of business.

Industry and product integration: Mining can be guided by industry and product family, ensuring that intent libraries reflect the organization’s unique landscape.

Sample-driven validation: Admins can now view sample cases and conversations that led to intent generation, increasing confidence in the mining process.

Autonomous and semi-autonomous learning loops

The July release introduces full and semi-autonomous learning loops. The Customer Intent Agent can continuously improve by extracting new intents and questions from completed conversations. The improved feedback loop, backed by usage data and dashboards, supports the authoring of high-ROI solutions and the creation of knowledge articles. The Customer Intent Agent closes the loop between customer interactions and organizational learning.

If your organization wants to reduce manual effort, improve resolution times, and deliver consistent customer experiences, now is the time to explore what this agent can do. Preview the Customer Intent Agent in your Dynamics 365 Customer Service environment today!