Claudeforce: Anthropic and Salesforce Fuse Claude Into the CRM

On August 26, 2026, Salesforce and Anthropic announced Claudeforce, a partnership that fuses Claude's reasoning directly into Salesforce's customer-relationship-management platform. The move builds on an already expanded relationship between the two companies: Claude's reasoning capability is combined with Salesforce's live data and workflows, so that AI agents can carry out a wide range of actions on a company's behalf without a seller having to leave the chat window to look anything up.

The first pillar, called "Salesforce in Claude" (Salesforce inside Claude), lets a seller simply type a prompt on Claude's own screen to run a situation analysis, handle a customer request, or draft a proposal — all grounded in Salesforce's actual records and workflows rather than in whatever Claude might otherwise guess. Even something as sensitive as offering a customer a discount stays bound by the rules a company has already defined inside Salesforce: Claude's suggestion cannot override them, only apply them.

37 built-in sales skills

The deployment available from day one, "Salesforce inside Claude," runs from within Claude and ships with 37 built-in sales skills that Salesforce and Anthropic designed together, spanning a seller's entire revenue cycle so that Claude can act on it directly rather than merely describe it. Crucially, every single action still routes back through Salesforce itself, which is what guarantees a company's business rules apply every time something gets executed, rather than being just a suggestion Claude happens to follow.

• Sales-meeting preparation
• Deal-health review
• Pipeline review

The #1 AI meets the #1 CRM. Welcome to Claudeforce. The Salesforce + Anthropic partnership to build products that put your trusted data to work inside Claude. 37 built-in sales skills, backed by 27 years of CRM experience. Connect once. Claude can reason across…

Benioff on the "AIforce harness" and "Headless 360"

Salesforce chairman and chief executive Marc Benioff said that, thanks to a new "AIforce harness" combined with "Headless 360," Claude now gets direct, governed access to Data 360, Tableau, Slack, and the rest of Salesforce's workflow, without a user ever needing to leave the chat window to fetch it. In his telling, Claudeforce fuses the probabilistic intelligence built into AI agents with the deterministic, trusted customer data and workflows that already live inside Salesforce.

Probabilistic models alone can't run a company. Deterministic systems alone can't reason. Claudeforce fuses Claude's reasoning with Salesforce's trusted data and controls. The AI is the interface.

A second deployment, "Claude inside Salesforce," reverses the arrangement and runs Claude within Salesforce's own trust perimeter instead. Salesforce and Anthropic say several such deployments are planned, with additional built-in skills covering other business functions, beyond sales, due to arrive later this year.

The announcement builds on ground Salesforce had already laid in April 2026, when it unveiled "Salesforce Headless 360," letting AI agents bypass Salesforce's own user interface entirely and reach every one of its functions directly through an API, a command line, or the Model Context Protocol — the same connective layer that now underpins Claudeforce.

Salesforce built its business, historically, around software meant to assist human sales representatives. Now, in an era when AI agents can automate a large share of sales and customer-support work, the company is trying to turn its accumulated customer data and its enterprise governance rules into a competitive edge against rival AI agents — precisely what Claudeforce is designed to demonstrate. Both companies have signaled that the partnership could later extend to AI providers beyond Anthropic.

Salesforce's own announcement, dated August 26, 2026, also carried a quote from Anthropic chief executive and co-founder Dario Amodei, framing the deal from Anthropic's side of the partnership.

Salesforce in Claude brings this same frontier intelligence into the systems where much of the world's commercial activity happens.

According to Salesforce, the capability is already live with select pilot customers, with an open beta planned for September 2026. The company also says its own employees have already logged 8.1 million hours of annualized productivity gains from AI tools this year, twice the figure recorded the previous quarter — a number Salesforce is using to argue that fusing Claude's reasoning with its own governed data is not just a product bet, but something it is already relying on internally.

For sales teams already running on Salesforce, Claudeforce lowers the bar for putting an AI agent to real work: the governance, the customer records, and the approval rules a company has spent years building into its CRM don't have to be rebuilt for an agent to use them safely. For companies still weighing whether to let AI agents touch live customer data at all, it is also a signal from two of the sector's biggest names that reasoning and deterministic business rules are meant to be paired, not left entirely to an agent's own judgment.