Microsoft just shipped 30 new CRM skills for Copilot in a single release cycle. That number is worth pausing on. Thirty discrete, named capabilities – covering customer context retrieval, deal summarization, service case actions, and more – dropped in one go, signaling that the AI agent race inside CRM platforms has moved well past the pilot phase. For sales and service teams, meaningful workflow changes are coming within months, not years.
If you’ve been tracking CRM news over the past quarter, you’ll have noticed the pattern: every major platform is racing to embed AI agents directly into the processes where work actually happens, rather than bolting a chatbot onto the side of an existing interface. That shift is consequential. It changes what RevOps teams need to measure, what sales reps actually touch during a deal, and how leaders think about the cost of running a CRM org at scale.
What Microsoft’s 30 New CRM Skills Actually Mean for Sales Teams
The core idea behind Microsoft’s latest Dynamics 365 update isn’t just “more AI features.” It’s a specific architectural bet: that AI agents in CRM should live inside the tools people already use – Teams, Outlook, Word – rather than requiring reps to log into a separate CRM screen to get customer context. Microsoft calls this “flow of work,” and it’s a meaningful distinction from how most CRM adoption has been sold for the past decade.
Historically, CRM adoption has always broken down at the same point. Reps don’t want to switch context. They’re mid-conversation, mid-email, mid-meeting, and opening another tab to log a note or pull an account summary creates enough friction that it simply doesn’t happen. The 30 new Copilot skills are designed to collapse that gap – a rep drafting a follow-up email in Outlook can now pull live account data, recent activity, and suggested next steps without leaving the compose window.
That has real implications for win rate tracking and pipeline hygiene, because better in-context tooling produces better data capture – which means better sales forecasting downstream. It’s a virtuous loop, when it works.
Why ServiceNow Is Attracting Salesforce Migrations Right Now
Work4Flow’s new 90-day migration program – moving CRM data, objects, and processes from Salesforce, HubSpot, and other platforms into ServiceNow CRM – is a specific market signal worth reading carefully. Ninety days is an aggressive timeline for a CRM migration of any complexity, and the program’s emphasis on “agentic context awareness intelligence” suggests the pitch isn’t just “cheaper alternative.” It’s “come for the cost, stay for the AI architecture.”
ServiceNow’s CRM offering has matured considerably. Its Configure Price Quote (CPQ) and Sales and Order Management modules are now substantial enough that enterprise buyers with complex quoting workflows have a credible reason to evaluate it. The migration program targets exactly those organizations – ones that built elaborate Salesforce orgs over years and are now asking whether the maintenance overhead is worth it.
For teams considering a move, the honest calculus involves more than licensing costs. Your sales pipeline configuration, custom objects, and historical activity data all carry migration risk. A 90-day program can work if the source org is reasonably clean – which is why Salesforce admin hygiene (more on that below) matters more than ever when you’re even thinking about switching platforms.
Goldman Sachs’ recent upgrade of Palantir to Buy with a $230 target, while Salesforce and ServiceNow held steady in the same trading session, reflects how the market is sorting enterprise software right now. The core platforms aren’t going anywhere – but the tooling built around them, and the migration services that move data between them, are becoming a meaningful competitive layer on their own.
The Quiet Problem That AI Can’t Fix: Org Hygiene
Here’s the uncomfortable truth about deploying AI agents in CRM: they’re only as useful as the data they’re reading. Garbage in, garbage out isn’t a new principle, but it’s more consequential now because AI agents will confidently act on bad data in ways a human rep might catch.
The Salesforce admin best practices conversation – which has resurfaced recently with new urgency – is directly connected to this. A well-structured org with consistent field naming, controlled permission sets, and regular data audits gives AI agents something reliable to work with. A sprawling org with duplicate records, unmaintained workflows, and twelve different ways to log a “meeting” will produce AI recommendations that are, at best, unhelpful and, at worst, actively misleading for a sales cycle in progress.
The practical checklist for admins preparing their orgs for agentic CRM tools looks something like this:
- Audit and merge duplicate account and contact records before enabling any AI summarization features
- Standardize picklist values and field labels across all objects – AI agents parse field names literally
- Review and retire inactive workflows, process builders, and flows that may conflict with new automation layers
- Establish a clear data ownership model so agents know which fields are authoritative
- Set up validation rules that prevent junk data entry at the source, not downstream
- Document your permission structure before agents start taking actions on behalf of users – scope creep in automation is a real risk
If you’re looking for more structured guidance, the CRM Guides section covers data governance and org setup in depth.
How AI Agents Change the GTM Motion – Not Just the Tech Stack
The more interesting shift isn’t technical. It’s strategic. When AI agents can reliably handle activity logging, account summarization, next-step suggestions, and even quote generation, the human work in a sales org moves up the value chain. Fast.
That has direct implications for how go-to-market teams are structured. If an agent can synthesize three months of email history and surface the two most relevant talking points before a call, you need fewer people doing research and more people doing judgment work – deciding which accounts to prioritize, how to position against a specific competitor, whether a deal fits your Ideal Customer Profile. Those are decisions that require context a machine can inform but probably shouldn’t own.
The teams that will get the most out of agentic CRM in the next 12 months are the ones that consciously redesign their workflows around the new capability rather than just turning on the features and hoping reps adopt them. That means defining which tasks the agent owns, which tasks the rep owns, and what the handoff looks like when the agent flags something that needs human judgment.
What the Deposyt-Square Partnership Tells Us About AI in SMB CRM
Not everything happening in CRM right now is enterprise-scale. The Deposyt and Square partnership – pairing Deposyt’s national agent network and merchant growth tools with Square’s payments platform for small businesses – is a useful reminder that the SMB segment has a completely different set of needs.
For a small business owner, the AI and CRM conversation is less about agentic pipeline management and more about whether the technology works without a dedicated admin. Deposyt’s approach – local human support paired with payments and CRM tooling – is a deliberate counter to the assumption that more automation always means less human touch. For some segments, a real person nearby who can actually fix the problem is worth more than a 24/7 AI agent that escalates everything.
That’s worth keeping in mind when evaluating the broader market. The agentic CRM wave is real, but it’s hitting enterprise and mid-market first. SMB adoption will follow a different curve, shaped more by simplicity and support than by feature depth. For a comparison of tools suited to different business sizes, the CRM Tools Directory is a useful starting point.
What Metrics Actually Change When AI Agents Enter the Picture
If you’re in a RevOps role, think carefully about which metrics get affected first. The obvious candidates are activity volume metrics – call logs, emails sent, notes added – which spike when AI assists with or automates the logging itself. That’s not necessarily a signal of better selling; it may just be better instrumentation.
The more meaningful metrics to watch are churn rate in accounts where AI-assisted service is active, changes in Customer Lifetime Value where agents are handling renewal touchpoints, and Net Revenue Retention trends over a 6-12 month window. Slower-moving, yes – but more honest indicators of whether agentic CRM is actually improving customer relationships or just making the database look tidier.
There’s also a Customer Acquisition Cost angle worth examining. If agents are handling qualification, follow-up, and some degree of discovery, the cost to move a prospect through the early funnel should decrease. Whether that saving shows up in the numbers depends heavily on how clean the underlying data is – which circles back to the org hygiene point above.
For ongoing coverage of how these metrics are evolving as AI agents become standard in enterprise CRM, the CRM Daily Newsletter covers new data and case studies weekly.
The Open Question Nobody Has Answered Yet
Thirty new CRM skills in Dynamics 365. A 90-day migration track from Salesforce to ServiceNow. Payments-and-CRM bundles for SMBs with local human support. These are all real developments moving in real time.
But the question that none of the vendors – Microsoft, ServiceNow, Salesforce, or anyone else – has cleanly answered yet is this: when an AI agent takes an action inside a CRM on behalf of a rep, and that action turns out to be wrong, who owns the error? Not legally, necessarily, but operationally. Does the rep catch it? Does the system flag it? Does a deal get mishandled before anyone notices?
The promise of AI agents in CRM is that they reduce the cognitive load on reps and surface better information faster. The risk is that they create a new kind of blind spot – automated confidence in data that hasn’t been questioned. As these tools move from opt-in features to default behavior, that’s the tradeoff that will define how much organizations actually trust what their CRM is telling them.