Dynamics 365 agents can now work inside almost any contact centre, not only Microsoft’s own. On 21 October 2026, Darya Mazandarany, Corporate Vice President for Dynamics 365 Contact Center, announced Microsoft Contact Center Intelligence: a set of purpose-built AI agents, quality scoring and analytics that sit on top of the telephony and routing platform a business already runs. An open connector reaches third-party platforms such as Five9, Cisco Webex Contact Center and Avaya, and the whole package is in global public preview today.
The opportunity here is bigger than it first looks. For years, the price of serious AI in customer service was a contact centre replacement project, with all the contract timing, retraining and risk that comes with it. Contact Center Intelligence removes that prerequisite. Service leaders can now add autonomous self-service, assisted resolution and full quality review to the platform their team already knows, and connect all of it to the customer record in Dynamics 365. The teams getting the most from it will be the ones who treat it as a fast pilot, not a long programme.
What is Microsoft Contact Center Intelligence?
Microsoft Contact Center Intelligence is a layer of AI agents and analytics that adds to an existing contact centre rather than replacing it. It launches with three agents. The Customer Assist Agent resolves customer needs on its own or helps a human representative in real time. The Quality Assurance Agent evaluates interactions against quality, compliance and brand standards. The Service Operations Agent, listed as coming soon, will help configure, monitor and optimise the contact centre and manage its costs. All three are built on the agentic capabilities of Dynamics 365 Contact Center and governed through Microsoft Agent 365.
Around those agents sits a practical toolkit. A new designer and customer conversation harness, started from Microsoft Copilot Studio, builds self-service across voice and digital channels. Interactions can be deterministic flows, generative conversations or a hybrid of both, and agents can run on a choice of models. Voice biometric authentication works during live calls and carries over when a call transfers to a person, through the contact centre or through Microsoft Teams. Grounding comes from Microsoft IQ, a company’s own knowledge sources and more than 1,700 connectors.
Why layering beats rip-and-replace
The usual answer to how to migrate CRM platforms, or contact centre platforms, is careful sequencing: assess, plan, cut over, stabilise. Contact Center Intelligence offers a welcome alternative for the service floor, because the migration no longer has to come first. The routing, queues and telephony stay where they are, and the intelligence arrives on top. That means a business can capture the benefit of AI agents this quarter and make any platform decision later, on its own timetable and with real performance data in hand.
The early customers show what that looks like. Siemens Healthineers plans to layer Contact Center Intelligence onto its existing Genesys and ServiceNow platforms to resolve routine and complex remote support without long queues. Thomas Jefferson University began the preview in September and expected to go live roughly six weeks later, bringing AI agents into its HR Service Desk for enrolment, eligibility, coverage and billing questions. TTEC, one of the largest customer experience outsourcers, describes the value simply as modernising without a rip-and-replace project. Six weeks from preview to live service is a pace most service leaders will find very encouraging.
Quality assurance on every conversation, not a sample
Most contact centres review a small sample of calls, because human quality review is expensive. The Quality Assurance Agent can evaluate up to 100 per cent of conversations instead, using the Multimodal Agent Score, the same standard Microsoft applies inside Dynamics 365 Contact Center. Custom guardrails cover the details that matter to a brand, such as greetings, compliance wording and sentiment. Today it evaluates historical self-service conversations, with real-time evaluation and coaching on the roadmap, so teams can build their scoring model now and switch it on live when that arrives.
This is also the most direct answer to how to measure CRM ROI in a service setting. Return on a CRM or service investment comes from measurable changes: faster resolution, fewer repeat contacts, higher first-contact resolution and better compliance. With every conversation scored, those numbers stop being estimates drawn from a sample and become a complete record. Leaders can compare self-service against human handling, see which knowledge articles actually resolve issues, and put a clear, defensible figure on what each improvement is worth. Full coverage turns quality from a periodic audit into a continuous source of insight.
Why is CRM data quality important for a service agent?
CRM data quality matters because an AI agent can only be as helpful as the information it is grounded in. When a Customer Assist Agent looks up an order, an entitlement or a previous case, it relies on accurate contacts, correctly linked accounts, current product data and well-maintained knowledge articles. Good data lets the agent resolve the issue in one conversation. It also keeps human handoffs smooth, because the representative receives the full context rather than starting again. Clean service data is the single biggest lever on how much value an agent delivers.
The encouraging part is that the work pays off twice. Tidying case categories, merging duplicate contacts and retiring outdated knowledge articles improves reporting and routing for human representatives straight away, before any agent goes live. Contact Center Intelligence then builds on that foundation through Microsoft IQ and its connectors, which can draw on systems beyond Dynamics 365 as well. A focused fortnight on the top twenty contact reasons and their supporting articles is often enough to give a first self-service agent a strong start.
How does this improve CRM adoption in the service team?
Improving CRM adoption rates usually comes down to one principle: make the CRM do work for people rather than ask work of them. Contact Center Intelligence moves in exactly that direction. The Customer Assist Agent can surface customer context and suggested answers while a representative is on the call, so the record becomes something that helps in the moment rather than something to update afterwards. When the system saves a representative time on every interaction, consistent use follows naturally.
Quality scoring helps here too, provided it is framed as coaching. When every conversation is evaluated, team leaders can recognise strong calls and share them, not only catch mistakes. Representatives see their own progress against clear, consistent standards, and new starters ramp faster with examples drawn from their own queue. Pairing the launch with short, practical training on how the assistant works, and inviting representatives to flag answers that need improving, turns the service team into active contributors to the agent’s success.
Agentforce vs Microsoft Copilot for CRM: how do the service routes compare?
Every major CRM vendor is now shipping service agents, which is good news for buyers. Salesforce offers Agentforce for Service across its Service Cloud estate, and HubSpot provides its Customer Agent inside Service Hub. Microsoft’s distinctive move with Contact Center Intelligence is the open connector: it is explicitly designed to sit on top of contact centre platforms from other vendors, and to govern the resulting agents in Agent 365 alongside the rest of a company’s Microsoft agents. Agent 365 already held nearly 40 million registered agents across tens of thousands of companies by July 2026.
The right choice depends on where a business already runs its customer records, its telephony and its productivity tools. A company on Dynamics 365 and Microsoft 365 with a Five9 or Avaya contact centre now has a particularly short route to production. Pricing is usage-based through Copilot Credits, so cost scales with adoption, although charges some contact centre vendors make for API or voice channel access sit outside that pricing and are worth confirming early. Comparing like for like on a real use case is the clearest way to decide.
What should a service leader do in the next 90 days?
The first month is for choosing the right starting point. Pick two or three high-volume, well-understood contact reasons, such as order status, password resets or appointment changes, and confirm that the data and knowledge behind them are in good shape. In parallel, check the connector route for your contact centre platform and any API or voice access costs with that vendor. Contact Center Intelligence is in global public preview everywhere except Austria, so most European teams can begin straight away.
The second and third months are for building and measuring. Configure a first self-service flow in the Copilot Studio harness, decide where it hands off to a person, and run the Quality Assurance Agent across past self-service conversations to set a baseline score. Then launch to a slice of traffic and compare resolution rates, handling time and customer sentiment against that baseline. By the end of the quarter you have evidence, a tested governance model in Agent 365 and a clear case for expanding to the next set of contact reasons.
The Sirocco perspective
We work with service teams on Salesforce, HubSpot and Dynamics 365, and we are pleased to see Microsoft make contact centre AI an addition rather than a replacement. Our clients rarely want to change their telephony and their CRM at the same time, and this release means they no longer have to. Because we do not sell licences, our recommendations start from the service process, the contact reasons and the data, and we will tell you plainly when a different platform’s agent fits better.
Our advice is to start with a narrow, measurable pilot and a short data tidy-up, then let the quality scores guide the next step. Teams that do this before the new year will enter 2027 with agents resolving real customer questions, every conversation measured, and a service floor that gets better each month. If you would like a second view on where to begin, schedule a consultation with our team.
Get in Touch
If you are weighing where AI agents fit alongside your existing contact centre platform and your Dynamics 365 records, tell us about your service set-up and we will help you map a first pilot that pays back quickly.
