Front page
NewsCRM & Data14m · 09:01 BST · 6 min read
AI Agents Trigger a New Battle for Customer Data Control
As AI agents begin to retrieve, interpret and act on customer information, the CRM and CDP market is shifting from data unification to governed decision-making. Oracle, Microsoft, Braze, SAP and Iterable are competing to control how customer intelligence is used.
I keep coming back to the same question when looking at the latest CRM and CDP announcements—who actually owns the customer when an AI agent is making the decisions?
The answer used to be relatively straightforward. CRM systems owned the relationship, while CDPs brought customer information together so marketers and service teams could understand and activate it.
AI agents are making that division much harder to sustain.
An agent handling a customer request may need a CRM record, a CDP profile, transaction history, consent information, service context, inventory data and business rules before it can decide what to do.
It’s creating a new competitive battleground for CRM and CDP vendors. Announcements from Oracle, Microsoft, Braze, SAP and Iterable this week indicate how quickly that battle is developing.
And the timing is particularly relevant. National Customer Service Week 2026is putting the spotlight on the people who will have to work alongside these systems as AI reshapes customer service.
National Customer Service Week Puts the Workforce in Focus
The CRM and CDP debate can sometimes become overly focused on platforms, data architecture and agents. National Customer Service Week is a useful reminder that there are people sitting at the end of all those systems.
Rhys Harris, AI Product Director at Content Guru, told CX Today that AI is likely to reshape contact center roles through augmentation and new responsibilities rather than wholesale replacement.
“AI is changing the workplace in much the same way earlier waves of automation have: boosting production and efficiency, while requiring workers to retool and develop new skills.”
The value of CRM and CDP data will increasingly depend on how effectively frontline employees can use the intelligence those platforms provide. Harris expects many contact center roles to evolve as AI assistants absorb routine, high-volume conversations, with experienced customer service professionals moving towards supervising AI performance.
“In the contact center environment, many roles and tasks are being augmented, rather than replaced. As AI assistants take on more of the routine, high-volume conversations, many frontline roles in the contact center will pivot toward supervising AI performance.”
That creates another reason for CRM and CDP vendors to think beyond the customer profile. If agents and employees are both making decisions with the help of customer data, the platforms need to provide context that is understandable, actionable and governed.
Harris also argues that organizations will need to develop new skills, while the relatively low barrier to entry of many AI tools should make adoption accessible to a wider group of employees.
“Adoption is largely a mindset shift, and the benefits become clear quickly once people engage with the tools directly.”
Harris sees a competitive risk in leaving employees behind as AI becomes part of everyday service work.
“For CX leaders, standing still is also a real risk: a human agent using AI will outperform, and ultimately displace, a human who chooses not to.”
That workforce transition makes the technology battle playing out across CRM and CDP platforms even more consequential.
Oracle Wants the Data To Come With Authority
Oracle made perhaps the boldest move with its launch of Fusion Claw on September 29, which it described as a governed execution runtime for its Fusion Agentic Applications, allowing AI agents to reason, adapt, re-plan and execute complex enterprise work.
Oracle launched 25 new Claw-powered applications spanning areas including sales, finance, HR and supply chain, taking its broader Fusion Agentic Applications portfolio to 75.
The interesting part for CRM buyers is the governance model. Oracle said that its agentic applications can work within defined permissions, policies, risk thresholds and decision rights. Its “Outcome Receipt” concept also provides an auditable record of the authority, evidence, decisions and actions behind an outcome.
The growing use of AI agents changes what customer data needs to mean. While an account record can tell an employee who the customer is and what has happened previously, an agent needs considerably more guidance, including what it is allowed to do, which information it can trust, which decisions require approval and what it must record after taking action.
Oracle is effectively trying to make those controls part of the application and data environment.
For Salesforce, Microsoft, SAP and the growing number of customer data platforms building agentic capabilities, that raises the question: Who gets to define the rules around customer data when the software starts acting on the customer’s behalf?
Microsoft Brings CRM Closer to the Live Customer Operation
Microsoft is approaching the problem from the contact center. On October 2, it announced real-time streaming analytics for Dynamics 365 Contact Center, giving supervisors live visibility into queues, conversations, workforce availability and service performance.
The system is designed to help supervisors spot operational problems and respond while they are happening, rather than waiting for historical reports to explain why service levels deteriorated.
That brings CRM closer to the operational machinery behind the customer experience and shows why customer data becomes more valuable when it is combined with real-time business context.
A service agent or AI agent may know a customer’s history, preferences and previous interactions but it becomes much more useful if it also knows that the relevant queue is overloaded, a particular product is unavailable or another service channel has capacity. The CRM record becomes one input into a live decision.
Microsoft is competing on something broader than customer relationship management—the ability to turn customer information into an operational response.
SAP Wants CRM Data To Understand What the Business Can Actually Deliver
SAP is approaching the same problem from the enterprise data layer. The company’s October 3 blog post discussing AI-ready customer data highlighted its Revenue Intelligence capabilities, bringing together sales, e-commerce, inventory, pricing and fulfilment information through SAP Business Data Cloud.
That is particularly relevant to the CDP debate, as although a unified customer profile is useful, an agent trying to make a recommendation needs to understand more than customer behavior. Customer data without operational context can produce an attractive recommendation that the business cannot fulfil.
SAP’s approach puts the CRM/CDP conversation inside a much wider enterprise data architecture. The implication is that customer intelligence is becoming inseparable from business intelligence.
In Brief
-
Braze offers another indication of where this is heading. The customer engagement platform announced Decisioning Studio Go, Agentic Standards and Operator Connect at its Forge conference in Las Vegas on September 29. Operator Connect allows users to access Braze capabilities through external AI tools including Claude, ChatGPT and Microsoft Copilot. The competitive question becomes less about which CRM interface wins and more about which platform can make its customer intelligence useful wherever the work takes place.
-
Iterable’s October 2 discussion of identity resolution might appear less dramatic than Oracle’s agentic announcement or Microsoft’s real-time analytics push, but it may actually be one of the most important pieces. When software can act autonomously, identity resolution, consent, permissions and customer context stop being separate CDP concerns and become controls around automated decision-making.
The CRM/CDP Battle Is Changing
The old distinction between the systems that store customer information and the systems that activate it becomes less useful when an agent needs to retrieve, interpret and act on information across both.
It also creates a fresh battleground between the major CRM vendors and specialist CDPs. For enterprise buyers, that means the next CRM or CDP evaluation needs to go beyond data unification. The customer record still matters, but the bigger prize may be control over what happens after the record has been understood.
rate this storyhelps rank stories across CX Today
Read nextordered by techtelligence · every pick explained
same beat · CRM & Data
Why Claudeforce Indicates CRM Complexity Is Breaking Customer Experience
23 Sept 2026same beat · CRM & DataOptimove AI Is Betting Big on Agentic Marketing. Most CRM Teams Aren’t Ready22 Sept 2026same beat · CRM & DataCRM’s Great Escape: Salesforce, HubSpot, Zeotap, and Zendesk Redraw the Data Battle21 Sept 2026
More from the CRM & Data desk
Interview
UJET CEO Vasili Triant explains why bolting AI onto CRM and ticketing stacks may add complexity instead of improving customer experience.
Nicole Willing · 23 Sept 2026Explainer
Optimove AI Is Betting Big on Agentic Marketing. Most CRM Teams Aren’t Ready
Agentic AI is moving into the CRM and marketing stack, Optimove AI is betting it has the controls businesses need.
Rebekah Carter · 22 Sept 2026News
CRM’s Great Escape: Salesforce, HubSpot, Zeotap, and Zendesk Redraw the Data Battle
This week’s CRM news covers Salesforce’s Dreamforce vision, HubSpot’s Smart CRM, Zeotap’s Open CDP, and Zendesk AI agents.
