Program report · filed November 19, 2026

CRM & Sales TechMeasured report

Microsoft bets contact center harness will outlast the AI model

Microsoft's Dynamics 365 contact center lead argues the durable AI layer is the harness around the model, and shows how a 20% discount decision slips outside CRM and the contact center.

By Marcus Bennett3 min read696 words

Program notes

  1. Microsoft's Darya Mazandarany used a 20% discount approval as the example of a customer decision made in meetings, emails, documents and Teams chats rather than in the CRM or contact center.
  2. Mazandarany says the durable layer is the harness around the models, not the models themselves, because models will keep turning over.
  3. Microsoft's stack combines Dynamics 365 Contact Center, Dynamics 365 applications, Microsoft 365 and Power Platform for enterprises already running Teams, Outlook, SharePoint and Dynamics 365.
  4. The interview offers no benchmarks, sample sizes or cycle-time measurements, so the benefit remains an architectural assertion.
  5. Contact centers have collected interaction data for decades through recorded calls, logged chats, categorized cases and measured feedback.
Better Together: Why CC + CRM + UC Unlocks Full AI Value
PlateBetter Together: Why CC + CRM + UC Unlocks Full AI Value — erwinbosman / pixabay

Microsoft's case for tying contact center, CRM and unified communications starts with a discount decision no system of record captured: whether to hand a customer an extra 20% off the deal.

Darya Mazandarany, Corporate Vice President of Dynamics 365 Contact Center at Microsoft, made the point in a CX Today interview: "Should we give the customer the extra 20% discount on this deal? That doesn't happen in the CRM or in the contact center. That's happening in meetings, emails, documents and Teams chats."

Her broader argument concerns what surrounds the model. AI models now arrive with stronger reasoning, lower latency, broader language support and more capable agentic behavior, faster than enterprise service roadmaps can absorb. "I think what's durable is the harness that the models are running in and the ability to give them the right tools and the right intelligence context. Models are going to come and go, but what's fueling them is the key."

What does the harness actually mean?

A capable model can summarize a call or draft a reply. Resolving the customer's problem requires identifying the account, understanding the open case, pulling the relevant policy, recognizing earlier decisions and acting inside the systems where work gets done. Microsoft packages that combination as contact center plus CRM plus unified communications (shared messaging, voice and presence across Teams, Outlook and SharePoint), and frames it as an operating model rather than a product bundle.

Where the customer decision actually happens

The representative on the call rarely holds the full answer, and the missing decision lands on a different team depending on the stage:

  • A retention offer at renewal needs approval from a sales leader.
  • A delayed delivery requires input from a supply chain team.
  • A disputed charge depends on a finance policy, or on an agreement reached earlier in an account meeting.

Those conversations live in email threads, Teams chats, shared documents and meetings the contact center cannot see. Representatives search for the answer, ask colleagues for help, or promise a callback, and that effort passes straight through to the customer.

What the stack changes for service teams

Microsoft's proposition is that Dynamics 365 Contact Center, the Dynamics 365 applications, the Microsoft 365 collaboration environment and Power Platform can pull business context closer to the point of service. For an enterprise already standardized on Teams, Outlook, SharePoint and Dynamics 365, that shortens the list of disconnected places a representative must search, and lets an AI agent ground an answer in more than a narrow contact center knowledge base.

Treat that as a vendor claim worth interrogating. The interview supplies no benchmarks, no sample sizes and no measured cycle times. The 20% example illustrates a workflow; it does not quantify an outcome. What Microsoft asserts is architectural. What it has yet to show is how much time a representative actually saves, or whether an agent can close a case without a human approval step.

What has to be true before agents act

Microsoft concedes the architecture needs boundaries. An organization must decide which collaboration data is relevant to a given customer interaction, who can access it, how long it is retained, and what an AI agent is permitted to do with it. That last decision usually sits jointly with sales leadership and service operations, because an agent able to read a retention thread can also act on one.

Service intelligence only counts when it changes something

Contact centers have logged interaction data for decades: calls recorded, chats captured, cases categorized and feedback measured. Microsoft's framing is that this history pays off only when it changes what happens next. The same test applies on the commercial side, where a recorded customer call or a categorized case earns its keep when it routes an approval, updates a deal record or changes what a representative does on the next touch.

As models keep arriving faster than service roadmaps can track, the durable question for service organizations shifts from which model to deploy to which system holds the context, the tools and the governance. Microsoft's bet is that the answer is the connected contact center, not the model itself.

— Marcus Bennett

via cxtoday.com (Source)

Filed under

Share this article:

More from Marcus Bennett

Marcus Bennett

Show full bio

Staff writer covering people, launches and day-to-day practice in Sales technique and methodology, sales enablement at Quota Brief.

120 articles

Also rated

  1. Five Acquisitions in Six Months Reshape Customer Service AI
  2. Microsoft Says Sellers Can Now Skip the CRM Entirely
  3. Legacy CRM’s 10% usage estimate sharpens the replacement debate
  4. CRM Vendors Tackle AI Costs and Customer Data Access
  5. Cisco and Epic Join Forces to Merge CCaaS and CRM Workflows

« Previous articleNext article »