| Highlights: |
|
A salesperson preparing for a customer meeting might normally open Salesforce, check the account, review recent activity, look through the sales pipeline, and piece together what has changed.
Salesforce and Anthropic want Claude to do more of that gathering for them.
On August 26, Salesforce announced Claudeforce, an expanded partnership with Anthropic that connects Claude more closely with Salesforce data, workflows, and business rules.
The first product is Salesforce in Claude, a plugin that lets employees work with Salesforce information directly from Claude. It's currently available to select pilot customers, with Salesforce planning an open beta in September 2026.
What is Salesforce in Claude?
The idea is to give Claude access to the business context that a normal AI conversation is missing.
If you ask a standalone AI assistant, "Which deals should I focus on today?" it knows nothing about your actual pipeline. With Salesforce connected, Claude could review the Salesforce information that employee is permitted to access, identifying deals that may need attention and helping the employee decide where to start.
Salesforce in Claude launches with 37 prebuilt skills for sales teams. Salesforce lists examples including meeting preparation, pipeline reviews, and deal-health reviews. The companies developed the skills together specifically for Claude, rather than simply connecting a general chatbot to a database.
Here, a "skill" means a task Claude has been set up to perform using Salesforce information and tools. For example, instead of asking Claude to explain how to prepare for a sales meeting, an employee could ask it to prepare for a particular meeting. Claude could then use information from the relevant account, pipeline, and connected systems to build the briefing.
Salesforce describes the plugin as being able to both read information and take certain actions. That could include updating Salesforce records or running existing workflows. Those actions still pass through Salesforce rather than giving Claude unrestricted control of the system.
Where Can People Use It?
The first version is focused on sales, and the examples Salesforce has published give a better picture of what the product could look like during a normal workday.
Preparing For a Customer Meeting
Claude could review an account before a call and bring together relevant information such as account history, recent activity, and current opportunities. Salesforce also describes skills for summarizing conversations and preparing for calls.
A salesperson might ask: "Give me a briefing for my 2 p.m. meeting with this account. What has changed since our last conversation?"
That’s a different type of AI task from generating a generic meeting checklist because the answer can be based on company information.
Understanding a Difficult Deal
Salesforce says Claude can analyze information connected to a deal, including calls, emails and changes in its stage, to help identify possible objections and suggest a plan for addressing them.
That could make it useful before a follow-up conversation where the salesperson needs to understand what has already happened rather than rereading the full account history.
Finding Renewals or Expansion Opportunities
Another use case highlighted by Salesforce is reviewing existing accounts for possible renewals or expansion opportunities. An employee might ask: "Which accounts have renewals coming up that I should pay attention to?"
Claude could then work from the Salesforce information available to that employee.
What About Company Data?
There are two separate questions worth distinguishing: what Claude is allowed to access, and how Anthropic uses the information sent to Claude.
For access, Salesforce says Salesforce in Claude follows the company's existing Salesforce permissions and business rules. The organization can also control how much freedom Claude has when taking actions.
For model training, Anthropic states that inputs and outputs from its commercial products, including Claude for Work, are not used to train its models by default. Organizations can choose to participate in specific programs where data is shared for model improvement, but that’s separate from the default commercial arrangement.
Salesforce in Claude also provides zero-data-retention options for supported Claude models. Anthropic has separately been developing additional enterprise controls called Enterprise Frontier Safeguards, which are intended to let eligible organizations keep certain monitoring data within cloud infrastructure they control. Those safeguards are rolling out separately and shouldn’t be treated as universally available today.
For companies evaluating Claudeforce, those details are worth reviewing against their own Salesforce agreement, Claude plan, and internal data policies rather than assuming every configuration works the same way.
Why Claudeforce is Interesting
The useful part of Claudeforce is the change in context. AI assistants like Claude can already summarize documents, draft emails, and analyze information that someone provides to them. Their usefulness becomes more specific when they can work with the systems a company already relies on.
A salesperson could move from asking: "How do I prepare for an important customer meeting?" to: "Prepare me for my meeting with this customer using our latest account information."
That sounds like a small change in wording, but it requires a very different connection behind the scenes.
Claudeforce is still at the pilot stage, and several of the broader capabilities Salesforce is showing are planned rather than generally available. The sales integration gives a clearer view of where the companies are heading: Claude working with business information and business processes directly, while Salesforce continues to control what information and actions each employee is allowed to use.
Learn more here: https://www.salesforce.com/news/press-releases/2026/08/26/salesforce-and-anthropic-announce-claudeforce/
© 2026 SVA Consulting