Salesforce’s Claudeforce plug-in makes AI agent rollouts look like a CRM upgrade

Salesforce used its Aug. 26 earnings release to make a very specific bet about how enterprises will buy AI next: as a CRM-native plug-in with prepackaged workflows, not as a separate model integration project.

Reuters reported that Salesforce expanded its partnership with Anthropic by rolling out a new plug-in that combines Salesforce with Anthropic’s Claude models. CNBC said the plug-in, called “Claudeforce,” can draft emails for sales reps, provide context, and update records through chat. The Wall Street Journal also reported Salesforce disclosed an expanded partnership that pairs its platform of business tools with Claude.

For operations, IT, and procurement leaders, the news isn’t the name of the plug-in. It’s what it implies: if “agents” arrive as packaged capabilities inside a system of record, the hard work shifts to data access, policy, and cost controls that live in day-to-day administration, not innovation labs.

Agent rollouts are starting to look like an admin and governance project

Salesforce’s Q2 numbers are part of why the company can push this integrated approach. Reuters reported fiscal second-quarter revenue grew 11% to $11.35 billion and that Salesforce raised its fiscal 2027 revenue outlook to $46.1 billion to $46.4 billion. CNBC reported the same 11% quarterly revenue growth figure and said adjusted EPS came in at $5.90 on $11.35 billion in revenue, with guidance for fiscal Q3 revenue of $11.42 billion to $11.50 billion.

Reuters linked the higher guidance to “continued momentum” in Agentforce, Data 360, and Slack. That trio is a useful clue for how Salesforce expects customers to implement agents: automate work in CRM, lean on a data layer that can reconcile and govern customer information, and surface the work inside collaboration.

CNBC pointed to a clear adoption metric: Agentforce AI products reached more than $1.5 billion in annualized revenue, a 240% increase year over year. That figure won’t predict any single enterprise’s ROI, but it does offer a reference point for the pace at which large organizations are moving to packaged agent products instead of building custom tools.

Claudeforce moves Claude model choice into day-to-day workflow design

CNBC’s description of Claudeforce is operationally telling because it includes both content generation (email drafting) and system actions (record updates). Once an AI tool can write back into CRM, the scope of risk and control changes. “Read-only copilots” live mostly in user training. “Write-capable agents” quickly become a controls and auditability discussion: what objects can be updated, when, under whose identity, with what approval flow.

The Wall Street Journal framed the partnership expansion as Salesforce combining its platform with Claude. Reuters characterized the plug-in launch as an expansion of a partnership first struck in June. The consistent message across the outlets is that Salesforce is not treating Claude as a loose integration, it’s being productized into the stack.

That productization tends to compress deployment timelines, but it also hardens vendor and platform choices. For organizations that already run Salesforce heavily across sales and service, this kind of plug-in can move faster than a horizontal AI platform rollout. For organizations with multiple CRMs or a fragmented customer data environment, the same plug-in could increase pressure to rationalize systems before agents can safely act.

The infrastructure echo: AI projects are turning into data and capacity contracts

The agent narrative is also playing out a layer below, inside the data platforms that must supply these systems. Diginomica reported Everpure’s fiscal Q2 2027 revenue rose 38% year over year to $1.19 billion, with operating profit up 77%, and that subscription services represented 42% of total revenue. It also reported that Storage-as-a-Service total contract value climbed 121% to $277 million, and that the company’s Evergreen//One offering reached an annualized TCV run rate above $1 billion.

Even though Everpure and Salesforce sit in different parts of the stack, the packaging trend rhymes. Diginomica described Everpure’s push into data intelligence, discovery, and contextualization to prepare enterprise data for AI. That’s the same work many CRM-driven agent deployments run into immediately: agents are only as useful as the data they can access and the policies that govern it, and enterprises increasingly want those capabilities delivered as managed services with predictable contract structures.

There’s a second operational parallel in Diginomica’s reporting: the push toward consumption and subscription constructs alongside hardware or core platform engines. Salesforce is doing it at the application layer with Agentforce and partner models; Everpure is doing it at the infrastructure layer with storage-as-a-service and data intelligence. In both cases, the buyer’s job becomes less about selecting a single product and more about defining limits: usage, workloads, data domains, and who pays when adoption scales.

What Salesforce-heavy enterprises should put into this quarter’s evaluation cycle

• Map the “write” surface area before pilots: If Claudeforce or Agentforce agents will update CRM records (as CNBC described), confirm which objects, fields, and workflows are in scope, and what human approvals are required for customer-facing actions.
• Ask for the bundle boundary in writing: Reuters said momentum in Agentforce, Data 360, and Slack is driving guidance. Clarify which capabilities are included in current SKUs, which are add-ons, and which are consumption-priced so forecasts don’t break after a successful pilot.
• Treat the model partnership as a support question: The Wall Street Journal and Reuters both described the Anthropic partnership expansion. For regulated or high-availability environments, confirm how incident response works across Salesforce and Anthropic dependencies, including escalation paths and change windows.
• Benchmark adoption with one number: Use CNBC’s Agentforce annualized revenue figure ($1.5 billion, up 240% year over year) as a maturity reference point internally when deciding whether to standardize on Salesforce-native agents or continue with departmental tools.
• If data is messy, don’t start in the noisiest channel: For organizations with fragmented customer records, start agent use cases where data ambiguity is lowest (internal case summarization, guided next steps) and expand to outbound email drafting and record updates after data governance work is complete.