Innovate Your Business
with Our Insights!

The new version allows people to ask questions like which deals are likely to close soon or who recently met with a specific customer. The system understands the question, looks across the CRM data, and returns an answer without the user needing to know how the data is stored or how to write queries. 

Microsoft says the change was needed because many CRM systems have grown complex over time. Companies customize their systems, add new fields, and connect multiple data sources. Earlier search tools often struggled with this complexity and failed when questions were unclear or crossed multiple tables. The updated Dataverse Search now thinks through the question step by step and adjusts its approach if the first attempt does not return the right result. 

The search experience also supports follow-up questions. If a user asks about open deals in one region and then asks a second question to narrow the list, the system remembers the earlier context instead of starting from scratch. It can also learn company-specific terms and abbreviations when administrators define them, helping it understand how each business talks about its own data. 

Microsoft tested the new system using real-world CRM scenarios and reported that it returns accurate results for most everyday questions, while still responding quickly enough for daily use. The update is already being used across Copilot experiences in Dynamics 365, Power Apps, and Microsoft 365. 

With this change, Microsoft is positioning Dataverse Search as a simple way for non-technical users to get answers from CRM data. The goal is to let people spend less time searching through screens and more time acting on the information they need. 

The update is designed to support organisations that generate revenue from renting equipment, tools, machinery, medical devices, technology assets, and other products rather than selling them outright. According to Microsoft, rental and leasing markets are expanding rapidly, with global revenues already exceeding $500 billion annually, while North American equipment and tool rentals alone are expected to cross $80 billion. 

Microsoft said many rental businesses today rely on disconnected systems for quoting, reservations, dispatch, billing, maintenance, and financial reporting. This fragmentation often leads to idle assets, slower deal cycles, billing errors, and poor customer experiences. The upcoming Dynamics 365 enhancements aim to bring these processes together inside a single ERP environment. 

Planned capabilities include rental quoting and reservations to confirm asset availability, flexible contract and pricing management for short-term, long-term, and rent-to-own models, inspection coordination at delivery and return, and billing and invoicing directly tied to rental activity. These features are intended to help organisations reduce manual work, improve accuracy, and speed up the end-to-end rental lifecycle. 

The rental functionality will be built natively across Dynamics 365 Finance, Dynamics 365 Supply Chain Management, Dynamics 365 Project Operations, and Dynamics 365 Field Service, allowing organisations to manage rentals as an extension of their core operations rather than as a separate system. 

The company also highlighted the role of its partner and ISV ecosystem, noting that the new rental foundation can be extended with industry-specific solutions and AI-driven capabilities using Microsoft Copilot Studio and agent-based automation. These extensions are intended to support specialised pricing, compliance, and asset lifecycle requirements across different rental-heavy industries. 

Fluent v2.0 is Microsoft’s latest design system and is already used across products like Teams, Outlook, and Windows 11. With this update, Customer Service now looks and works the same way as the rest of the Microsoft ecosystem. Screens are cleaner, navigation is simpler, icons are clearer, and layouts are easier to understand. The goal is straightforward: help users do their work faster with less effort and confusion. 

The refreshed design touches every part of Customer Service. Copilot interactions are easier to use, inboxes and conversations are simpler to scan, and case management screens are cleaner and more organized.  

Timelines load faster and are easier to read, scripts and macros are clearer to follow, and voice and chat controls are more intuitive during live interactions. Knowledge articles, email editors, SLA tracking, and Smart Assist recommendations have all been updated to improve visibility and usability. Supervisor dashboards and Power BI reports now present data more clearly, making it easier to track performance and take action. 

Microsoft says the update directly addresses long-standing feedback about outdated interfaces and inconsistent layouts. Earlier versions required users to adapt to different designs across screens, which slowed down work and increased training time. Fluent v2.0 removes that friction by making every part of Customer Service feel familiar and predictable. 

The redesign also improves accessibility, with larger controls, better spacing, and support for WCAG 2.1 standards, ensuring the platform is usable for a wider range of users. Microsoft reports early results showing higher feature usage, improved satisfaction scores, and positive feedback from service teams and customers alike. 

The benchmark evaluates autonomous sales agents across three core jobs required for qualification: lead research, personalized outreach, and multi-turn engagement. In tests completed on December 4, 2025, Microsoft reported that Sales Qualification Agent scored higher than ChatGPT across all three dimensions, with a 6 percent improvement in research quality, a 20 percent improvement in outreach effectiveness, and a 16 percent improvement in engagement quality. The evaluations were conducted using identical lead datasets, knowledge sources, and contextual inputs, with outputs scored independently by human reviewers and an LLM judge built on GPT-5.1. 

According to Microsoft, the benchmark reflects months of internal and customer-partnered testing that began during the agent’s public preview earlier in 2025. The evaluation dataset consisted of 300 synthetic leads modeled across three industries and included only the types of attributes typically available to sales teams before enrichment, such as company name, email, and limited background context. Both systems were assessed on accuracy, relevance, completeness, grounding, and business-specific success criteria, including personalization quality, intent discovery, and timing of handoff to human sellers. 

Microsoft attributed the stronger performance of Sales Qualification Agent to its sales-specific design, including the use of agentic Retrieval Augmented Generation to iteratively research leads, deeper contextual awareness of what a seller offers, and tighter integration with organizational knowledge sources such as CRM records and SharePoint. Unlike ChatGPT, the agent also operates autonomously, allowing it to research, engage, and disposition leads without continuous seller input, a capability Microsoft says materially reduces time and cost required to build qualified pipeline. 

By formalizing the Sales Qualification Bench, Microsoft is signaling a broader push toward transparent, repeatable evaluation of AI agents in revenue-critical workflows. The company stated that additional benchmarks covering future sales agent capabilities will be added to the Microsoft Sales Bench collection, positioning agentic CRM not as a conceptual advancement but as a measurable, production-ready shift in how enterprise sales organizations qualify and convert demand. 

Announced by Jared Spataro, the release integrates OpenAI’s latest model directly into Microsoft 365 Copilot and Microsoft Copilot Studio, giving users access to two modes: GPT-5.2 Thinking for complex problem-solving and strategic analysis, and GPT-5.2 Instant for faster tasks such as writing, translation, and skill-building. Users can select GPT-5.2 directly from the model menu in Copilot Chat and Copilot Studio. 

In Microsoft 365 Copilot, GPT-5.2 connects with Work IQ to reason across meetings, emails, documents, and organizational data, enabling tasks such as extracting strategic insights from conversations, preparing for meetings based on prior interactions, and connecting discussion outcomes to business objectives and milestones. Microsoft says this combination is designed to make AI more context-aware and practical for real-world work scenarios. 

The update also advances Microsoft’s broader strategy of offering model choice within Copilot, allowing organizations to use the best model for specific tasks while maintaining enterprise-grade security, privacy, and compliance. In Copilot Studio, GPT-5.2 is available today in early release environments, and existing agents running on GPT-5.1 will automatically transition to the newer model. 

Microsoft described the rollout as part of its ongoing effort to “build the future of work together,” noting that feedback from customers will continue to shape how GPT-5.2 is refined and deployed.