Microsoft is adding a new layer of AI capability to Dynamics 365, expanding its family of “business process agents” across sales, service, and contact center operations to accelerate decision-making and enhance customer experience across industries. The updates reflect the company’s broader plan to make Copilot the user interface for enterprise AI.
The updates include new agents for sales, service, finance, and supply chain management, each designed to automate key tasks and provide real-time insights across business functions.
The latest release includes the Quality Evaluation Agent in Dynamics 365 Customer Service and Dynamics 365 Contact Center, which will become available beginning October 24 and which gives supervisors a real-time view of service performance across both human and AI-led interactions. Rather than sampling a handful of cases, it uses AI to review the majority of cases and conversations, flag anomalies, and trigger corrective action for faster and more consistent quality management.
Also moving to general availability on October 24 are the Case Management Agent in Dynamics 365 Customer Service and the Customer Knowledge Management Agent and Customer Intent Agent in Dynamics 365 Customer Service and Contact Center. These are all designed to simplify case routing, knowledge sharing, and customer understanding.
In Dynamics 365 Field Service, the Scheduling Operations Agent, which will be available in public preview, is designed to keep schedules agile and service running smoothly.
These tools are part of Microsoft’s ongoing push to embed intelligence across its business applications, to help enterprises move from reactive service to proactive engagement.
Copilot Studio as the AI Interface for Enterprise
At Microsoft’s recent AI Day, Rob Smithson, UK AI Business Processes Lead at the tech giant, described Copilot as the UI for AI. Within the enterprise, teams will increasingly interact with tools like Dynamics 365, ERP, and CRM systems through Copilot interfaces in the future—either via chat, in-app prompts, or embedded automation. Copilot can manage the background tasks, drafting case summaries, suggesting actions, or analyzing conversation tone, so that employees can focus on creative problem-solving and relationship-building.
“Those different programs provide data back to the interface so users can ask questions or take actions in the flow of work.”
To support customization Copilot Studio extends this further, acting as what Microsoft calls the “extensibility layer.” It lets organizations design and customize their own agents for their specific systems and needs, connecting securely to enterprise data without extensive coding or reliance on third-party systems.
Dynamics 365 seamlessly connects with Power Platform and Copilot Studio. The result is a unified ecosystem where apps, workflows, and data connect through Microsoft’s AI foundation.
The Dynamic 365 agents run on Microsoft Dataverse, giving teams a unified data foundation across sales, service, and finance. It’s the architecture powering use cases like Banco PAN, which is using Dataverse as a core part of its Dynamics 365 system to enable real-time integration.
“Our operators now have immediate access to the customer’s history and can resolve issues more quickly,” said Tulio Prado, Service Superintendent at Banco PAN.
To help customers evaluate agent performance, Microsoft has also introduced Sales Research Bench, a standardized framework that scores AI models across metrics such as accuracy, relevance, and clarity. Early results showed Microsoft’s own Sales Research Agent outperforming ChatGPT-5 and Anthropic’s Claude Sonnet 4.5 in composite quality scores.
The new capabilities highlight Microsoft’s steady march toward what it calls an “agent-led, human-directed enterprise.”
Nonprofit organization yourtown (Kids Helpline) has found that Dynamics 365 agents are helping enhance its teams responses. Helen Vahdat, Chief Information Officer at yourtown said:
“By adopting agents in Dynamics 365 service solutions, we’re making every interaction faster and more empathetic. In a service where demand exceeds capacity, this can be a game changer.”
“In our fundraising unit, we’re also exploring how agents can manage inbound calls to reduce abandonment rates from 20 to 30% to under 5%—directly lifting revenue streams that fund vital services.”
Enticing Enterprise Experimentation
But despite the hype around AI agents, it’s becoming clear that achieving enterprise adoption at scale is a challenge for vendors. An indication of this is Zoho adding free agentic AI tools to its apps to encourage uptake. It’s one of several software vendors making agents free, or nearly free, to get customers experimenting.
Microsoft is also taking that step, “making it easier to get started with agents” by including 1,000 Copilot Credits per user, per month, pooled at the tenant level, for Dynamics 365 Premium products, beginning in late November. That includes Dynamics 365 Customer Service Premium, Sales Premium, Supply Chain Management Premium, and Finance Premium.
Customers can use the credits to try out agents in scenarios relevant to their business. Customers can buy additional Copilot Credits as needed.
To increase transparency into AI performance, Microsoft is also launching standardized AI benchmarks—starting with the Sales Research Bench, which measures the performance of agents across metrics such as accuracy, relevance, and clarity.
“We also plan to publish agent performance regularly to reduce friction and help leaders make confident, data-driven decisions,” Bryan Goode, Microsoft’s Corporate Vice President, Business Applications and Agents, wrote in a blog post.
Early testing found that the Sales Research Agent in Dynamics 365 outperformed OpenAI’s ChatGPT-5 and Anthropic’s Claude Sonnet 4.5 on composite quality scores, with plans to make the benchmark dataset public in the coming months.
As agentic capabilities expand across Microsoft’s business applications, the company’s AI strategy is coming into sharper focus: unify data, integrate automation, and bring AI into the flow of every customer interaction.
