CRM has traditionally been a place where employees go to update records, check account history, and find the information they need.
But is that still the case?
The biggest CRM announcements from the past seven days point to a model where data, business rules, and customer context are being pushed out of the conventional CRM interface and into AI agents, collaboration tools, data clouds, and service workflows.
Salesforce chose Dreamforce to set out its new CRM vision… which involved spending less time in your CRM. HubSpot, on the other hand, wants to remove manual CRM upkeep altogether. Zeotap is challenging the idea that a CDP vendor should host customer data in the first place. Zendesk, meanwhile, is betting that specialized AI agents will need to work across the systems enterprises already use.
As Marc Benioff, Chair and CEO of Salesforce, put it: “AIforce is here to unlock trapped value from anywhere.”
Here’s what that means for CRM and customer data leaders.
Salesforce Wants CRM to Work Beyond Salesforce
Dreamforce 2026 was dominated by Salesforce’s effort to turn CRM into an invisible but tightly governed enterprise layer.
The vendor unveiled AIforce, its umbrella term for agentic experiences that bring Salesforce context, permissions, workflows, and actions into external interfaces such as Slack, Microsoft Teams, and Anthropic’s Claude.
At the center of the strategy is ClaudeForce, which allows employees to access Salesforce customer data and take authorized actions from within Claude Enterprise.
Salesforce also introduced Slackforce, alongside Slackforce Surfaces, which turns prompts into live reports, dashboards, and working views based on Salesforce and connected enterprise data.
Parker Harris, Co-Founder of Salesforce, summed up the company’s ambition when he said: “You may never log in to Salesforce again.”
The vendor was quick to point out that the new focus does not mean Salesforce is becoming less important.
Instead, Salesforce is positioning itself as the governed system underneath AI interactions. It wants customer records, metadata, business rules, and identity controls to travel with employees and AI agents wherever they work.
In practice, this means that if a service employee can check an entitlement, update a CRM record, or trigger a follow-up action without switching between systems, CRM becomes less of an administrative burden.
The tradeoff is that the governance burden gets heavier. As CRM actions move into new interfaces, leaders will need to know exactly which customer data an AI agent can see, what it can change, and how those changes are tracked.
HubSpot Makes a Case for the Self-Updating CRM
HubSpot has taken a more practical approach to the same underlying problem of CRM data often being incomplete because people do not keep it up to date.
At its Fall 2026 Spotlight, the company introduced a self-updating Smart CRM that automatically captures and synchronizes calls, emails, and meetings. HubSpot says the goal is to give sales, marketing, service teams, and AI agents a current view of the customer without relying on manual data entry.
It also launched Context Home, a centralized place for teams to view and manage the business context powering the platform. That includes a Context Score, intended to show where customer and business data is incomplete.
In discussing the news, Duncan Lennox, Chief Product and Technology Officer at HubSpot, said:
“We hear from customers every day that they don’t want to think about which AI tools to use, they just want outcomes.”
HubSpot’s argument appears to be that AI agents are only as useful as the CRM data beneath them. If the latest customer conversation, contact details, or buying signals are missing, an agent may produce an answer quickly, but it may still be the wrong answer.
The challenge now is whether HubSpot can make “touchless” CRM data reliable enough for customers to trust it. Automatic capture is helpful; accurate customer context is the harder job.
Zeotap’s Open CDP Puts Data Sovereignty on the Table
Zeotap has launched Open CDP, a composable customer data platform designed to run on open-source compute and the customer’s own lakehouse.
That may sound like infrastructure news, but it gets to the heart of a growing CRM and customer data dispute: where should customer data live, and who should control the compute used to process it?
Open CDP uses Apache Spark for identity resolution, audience computation, and activation, while storing data in Apache Iceberg tables on the customer’s own object storage. It also supports identity graphs, segmentation, reverse-ETL activation, consent governance, and journey orchestration.
Projjol Banerjea, Founder and Chief Product Officer at Zeotap, said:
“Customers keep their data, their storage, and their catalog; we bring the open engine and the CDP intelligence on top.”
For regulated businesses, that will be an appealing proposition. Data residency, cost control, and vendor lock-in are becoming central concerns as customer data volumes rise and AI systems need more real-time context.
Zeotap is effectively arguing that composability has not gone far enough. The CDP should be able to sit on top of a customer’s existing data estate, rather than becoming another proprietary place where customer information is copied and controlled.
Zendesk Says Generic AI Agents Are No Longer Enough
Zendesk has introduced Specialized AI Agents, including commerce-focused Industry Agents and configurable Custom Agents.
The launch is primarily a customer service announcement, but it has clear CRM implications. These agents are designed to combine company knowledge, policies, workflows, and connected systems so they can take action across a customer journey, rather than simply answer isolated questions.
Zendesk’s commerce agents can connect with Shopify, Narvar, Stripe, and Riskified to support work such as returns, refunds, delivery queries, and exchanges. The company says its Custom Agents can also run proactively when a business event occurs.
Indeed, Shashi Upadhyay, President of Product, Engineering, and AI at Zendesk, made the bold claim that “the era of generic, one-size-fits-all AI agents is over.”
He may be right. Yet specialized AI is only as effective as the data it can access and the guardrails placed around it.
That is the common thread across this week’s CRM news. Whether it is Salesforce moving CRM into AI interfaces, HubSpot updating customer context automatically, Zeotap keeping data in a customer-controlled lakehouse, or Zendesk giving agents access to business systems, CRM is becoming less about just storing information.
The real battle is over who can make that information useful, current, and safe enough for AI to act on it.