Quick Answer:
Dynamics 365 Copilot in 2026 has evolved from a chat assistant into an agentic AI layer across various platforms including Sales, Finance, Supply Chain, Customer Service, and Business Central. Key developments include Work IQ integration with Dataverse, custom agent building, multi-model intelligence, and new licensing changes effective July 1, 2026.
If you've been tracking Microsoft's AI ambitions over the past two years, you already know that Copilot is everywhere.
Copilot is now embedded in Word, Excel, Teams, and Outlook; increasingly, it's woven into the Dynamics 365 applications that run core business operations. Now, nearly a third of the way through 2026, the Dynamics 365 Copilot story is changing in ways that matter far more than another round of incremental feature updates.
The shift is architectural. Microsoft is moving Copilot from a chat-based assistant that answers questions toward something more ambitious: AI agents that can reason over your ERP and CRM data, break complex requests into actionable steps, and take action on those steps within defined guardrails. The agentic AI shift is a fundamentally different proposition than "summarize this email thread," and it carries different implications for how organizations plan, implement, and govern their Dynamics 365 environments.
In this guide, we’ll break down:
- What’s actually new
- What it means for your business
- Where the practical pitfalls are hiding
- What it means for your business
We'll cover Copilot's capabilities across each major Dynamics 365 application, walk through licensing and pricing changes, and address the governance and data readiness work that determines whether Copilot delivers real value or just adds noise.
How Copilot Has Gone from Sidecar to Agent
The original Copilot experience in Dynamics 365 Finance & Operations, which appeared in early 2023, was essentially a contextual help panel. Users could ask questions like "How do I create a purchase order?" and receive step-by-step guidance. It was certainly useful, but closer to an interactive how-to guide or manual than an intelligent assistant.
Since then, the trajectory has been steep. By mid-2025, Copilot in Finance & Operations could interact with live data, apply predictive models, and automate routine processes. The 2025 Release Wave 2, which rolled out between October 2025 and March 2026, brought hundreds of new capabilities across various Dynamics verticals, like Sales, Customer Service, Contact Center, Field Service, Finance, Supply Chain Management, Project Operations, Human Resources, Commerce, and Business Central.
The story of Copilot in 2026, however, centers not on chatbots, nor macros, nor scripted workflows, but on agents.
Agents in the Dynamics 365 context are AI-powered assistants that interpret high-level business goals written in natural language and translate them into structured, multi-step ERP and CRM actions. Traditional automation says "if X happens, do Y." An agent says "here's the goal I was given. I will figure out the steps, execute within boundaries, and ask for approval where needed."
That distinction matters enormously for organizations running complex operations, and it's the lens through which everything else in this guide should be read.
What Copilot Does Across Dynamics 365 Applications
Copilot's capabilities vary by application, but the underlying pattern is consistent: It is a completely native embedded AI that surfaces insights, drafts content, automates repetitive tasks, and, increasingly, takes autonomous action within the guardrails it’s given.
Sales
Copilot in Dynamics 365 Sales has become one of the more mature implementations. Sellers get AI-generated meeting preparation summaries that pull from previous meetings, recent notes, and CRM data. During Microsoft Teams calls, Copilot offers real-time tips and suggested responses when customers mention competitors or raise objections. After calls, it generates summaries, highlights action items, and drafts follow-up emails grounded in the conversation context.
The newer agent capabilities go further. The Sales Qualification Agent, Sales Close Agent, and Sales Research Agent can work semi-autonomously. They can take on the grunt work of qualifying leads, identifying next-best actions, and enriching account data without a human initiating each step. These agents consume Copilot Credits on a pay-as-you-go basis, which we'll cover in the licensing section.
Finance
Finance teams see some of the highest-leverage Copilot use cases. The AI can predict which customers are likely to delay payment, prepare collection reminders in natural human-like language, and reconcile transactions against bank statements.
On the collections coordinator workspace, Copilot generates AI-powered summaries that reduce the time spent reviewing customer details. This may be a small thing per account, but significant when multiplied across hundreds of open balances, every pay period of every year. That’s real person-hours being saved, which is a boon for your team (with much less frustrating busywork).
The 2026 Release Wave is adding agents designed to accelerate the financial close process and automate reconciliation tasks. These are financially consequential workflows, so the agent runtime enforces approval gates and exception handling to keep accountability where it belongs.
Supply Chain Management
Copilot's proactive side really shows when it comes to supply chain management. Rather than waiting for a planner to ask the right question, Copilot monitors for external risk signals, like weather events, supplier financial instability, or geopolitical disruptions, and flags them before they cascade into missed shipments or production stalls. It analyzes supplier performance and order histories to predict delays, then drafts communications to affected partners so your team can focus on mitigation rather than email.
The Supplier Communications Agent automates routine vendor interactions like order confirmations, schedule change notifications, and document requests. For 2026, Microsoft is adding demand planning that accounts for promotional events and seasonal patterns, quality management improvements for sample handling, and warehouse app upgrades. These aren't necessarily headline features, but they address the kind of operational friction that compounds fast across a large supply chain.
Customer Service
Service reps can now open their day in Copilot with a prioritized case summary, review and update D365 records, and draft responses grounded in case history and knowledge base articles, all from within the same app.
Planned for early April 2026 is a bigger bombshell of a development. This is when Customer Service gets Work IQ integration with Microsoft 365 Copilot (more on Work IQ below). The short version: Reps will pull D365 case data into the M365 Copilot interface alongside email, calendar, and Teams conversations, collapsing the context-switching that eats into resolution times.
Business Central
If you're an SMB running Business Central, the 2026 updates deserve close attention. The new AI Development Toolkit lets consultants, developers, and power users build custom agents directly inside Business Central – not by using traditional AL code, but by describing goals in natural language, testing in sandbox environments, and reviewing execution logs that show the agent's reasoning at each step. Agent definitions export as JSON and can be version-controlled in GitHub, and publishers can use Microsoft-managed Azure OpenAI resources by default, removing the need for customers to stand up their own AI infrastructure.
Project Operations
Project Operations gets a lighter Copilot touch, but a very practical one: AI-generated status reports that combine scheduling and financial data, continuous risk scanning for patterns like recurring delays or budget creep, and improved mobile experiences for time and expense entry.
Work IQ and M365/Copilot Integration
One of the most consequential 2026 developments isn't a feature within any single Dynamics 365 app; rather, it's the integration layer connecting them all.
“Work IQ” is Microsoft's name for the intelligence framework that pipes Dataverse business data into the broader Microsoft 365 Copilot experience. In practical terms, users can interact with Dynamics 365 Sales, Customer Service, and Power Apps data conversationally through the M365 Copilot interface, without opening the D365 application itself. An HR team can query a custom Power Apps application from within Copilot; a service rep can review priority cases and update records without leaving Outlook or Teams.
Public preview for Power Apps launched in March 2026, with Dynamics 365 Sales and Customer Service following in early April. Microsoft is also onboarding partner applications like Adobe Express, Box, Canva, Figma, and Monday.com through the Microsoft 365 Agent Store.
The implication worth sitting with is that Microsoft is positioning M365 Copilot as the primary interface layer for business applications, with Dynamics 365 apps serving as data and logic engines underneath. If you're planning your D365 strategy, that convergence should factor into your training and change management decisions now, not after the rollout.
Licensing and Pricing for Microsoft Copilot: What You Need to Know
Copilot in Dynamics 365 doesn't follow a single pricing model, so it's worth breaking this down layer by layer.
At time of writing, the Microsoft Copilot pricing breakdown looks like this:
Microsoft 365 Copilot runs $30/user/month for enterprise and $18/user/month for SMBs (under 300 users) through the Copilot Business SKU. This license covers Copilot in the M365 productivity apps and unlocks the Work IQ integration with Dynamics 365. If you're already on Dynamics 365 Sales Premium, adding M365 Copilot automatically enables the full Sales in M365 Copilot capabilities with no separate add-on.
Dynamics 365 app licenses include basic embedded Copilot features, like record summarization, contextual help, and AI-assisted drafting, at no additional cost. The more advanced agent capabilities, however, consume Copilot Credits.
Copilot Credits are available through pay-as-you-go billing or prepaid Copilot Credit Commit Units, which are cheaper at the tradeoff of having to pay in advance. Dynamics 365 Premium SKUs include 1,000 credits per user per month, pooled at the tenant level.
Autonomous agent triggers cost 25 credits each ($0.25), so if you have 50 users on Premium licenses, that's a pool of 50,000 credits per month, which translates to 2,000 autonomous agent actions before you start paying overage. Whether that's plenty or tight depends entirely on how aggressively you're using agents, but for most organizations dipping their toes in, the included allotment should cover early adoption comfortably. Still, it’s worth modeling expected agent usage before you commit to this.
One more thing to budget for: Microsoft 365 suite price increases take effect July 1, 2026. E3 will go from $36 to $39/user/month, E5 from $57 to $60, Business Basic from $6 to $7, and Business Standard from $12.50 to $14. If you're evaluating Copilot ROI, calculate against the combined licensing picture rather than looking at Copilot costs in isolation. And if you're planning to renew, doing so before July locks in the current rates.
Governance and Data Readiness
This is the section most organizations skip until it's far too late.
Copilot doesn't necessarily create new security risks; what it does is amplify existing ones. If your SharePoint permissions are overly broad, Copilot will surface sensitive files to users who technically have access but probably shouldn't. If your Dataverse records have inconsistent classification, agents will reason over messy data and produce messy outputs.
In other words: The AI is only as trustworthy as the environment it operates in.
Microsoft has built a governance framework called the Copilot Control System, organized around three priorities:
- Safeguarding data
- Preventing misuse
- Ensuring compliance
The core tools are Microsoft Purview, which handles sensitivity labeling, DLP policies for Copilot interactions, and insider risk alerts, and SharePoint Advanced Management, which provides data access governance reports and lets you send access reviews to site owners.
What a Governance-Ready Deployment Looks Like
In practice, this comes down to a handful of things you should probably be doing anyway:
- Audit your SharePoint and OneDrive permissions and identify sites with organization-wide access that shouldn't have it.
- Implement sensitivity labels. At Microsoft's own internal deployment, every container must be labeled or it gets deleted. That's aggressive, but the principle is sound: labels propagate from containers to files, and Copilot respects them.
- Configure DLP policies that prevent Copilot from processing sensitive files in its responses.
- Clean up stale content, including inactive sites, orphaned Teams channels, and outdated OneDrive files. Less noise means better signal.
- Use Purview's Data Security Posture Management for AI to monitor prompts, responses, and referenced files as your audit trail.
The most important thing, though, is to start. Roll out to a pilot group, measure adoption through the Copilot Dashboard, and refine your policies before scaling. Waiting for perfect governance is counterproductive, but piloting without visibility into what Copilot is accessing is reckless. The organizations getting the most from Copilot are treating it as a data governance project that happens to involve AI, not the other way around.
Microsoft Copilot Frequently Asked Questions
