Invest2 distinct publishers2 min readPublished
Thirty-seven prebuilt sales skills put Salesforce data and rules inside Claude. Salesforce keeps enforcement and the system of record; Anthropic gets the interface sellers actually sit in.
The Investor · Invest desk

Compiled by The InvestorSomething wrong?How this is made
The asset Salesforce keeps is narrower than the announcement language suggests. According to the company's release, every action taken from the plugin routes back through Salesforce so business rules are enforced when the action is taken [5]. Reading a customer record is the commodity half of that arrangement, and any model vendor gets there eventually. The enforcement point is the expensive half, because it is not really software: it is an accumulation of one company's rules about who may discount what and who signs off.
The naming tells you who needed the deal more, and so does the tape. Salesforce is down about 22% this year after falling 20% in 2025, while the Nasdaq is up roughly 35% since the end of 2024 [9]. Compounded, that is a drawdown of about 38% against an index that gained a third, a gap of some 73 percentage points [10]. On the other side, Anthropic told investors its annualized revenue run rate reached $65 billion at the end of July, seven times the year before [13], which puts the year-ago figure near $9 billion [14]. The extended-hours move is not a clean verdict on Claudeforce, since it arrived with the quarter [12], and it leaves the stock roughly 13% below where it started the year [15].
The framing is the part worth reading twice. Salesforce's own release argues that for decades enterprise software made people navigate static UI by hand, and that software is now a system powering every interface [7]. Benioff's phrasing is that the UI is the AI, and that probabilistic intelligence alone does not run a company while deterministic systems do not reason [16]. That is a CRM vendor describing its own screen as overhead. The screen was also where seats got justified at renewal, and where adjacent products got introduced.
What actually ships is smaller than the rhetoric. The skills cover meeting prep, deal health review and pipeline review, packaged as an AI CRO for sellers [2], plus composing emails and updating records [3]. Benioff says this is how every business will run [16]; for now it is how a pilot group runs [11]. The second half of the partnership, Claude inside Salesforce [17], is the half that keeps sellers on Salesforce's own surface, and it is the half that arrived without a number attached. Amodei's version of the value is that companies can point Claude at the customer context they have been building in Salesforce for decades and use it to run their businesses [20]. Pointing is the operative verb. The party holding the pointer is not the party being pointed at.
Ranked by verification strength, evidence, and original report placement.
Salesforce and Anthropic announced Claudeforce, an expanded strategic partnership, launching with Salesforce in Claude, a plugin with 37 prebuilt sales skills that let sellers and agents reason over live revenue context, automate pipeline updates and take governed action from within Claude.
The 37 prebuilt sales skills include meeting prep, deal health review and pipeline review, and Salesforce describes the plugin as giving sellers an AI CRO that understands the entire revenue cycle.
The plugin's 37 pre-built sales skills allow Claude users to compose emails, update records and take other relevant actions.
Salesforce says Salesforce in Claude routes actions through Salesforce to help ensure business rules are always enforced when an action is taken.
Claudeforce is available to a select group of pilot customers and will launch in a preview capacity next month, with additional pre-built skills available later this year.
Marc Benioff said the UI is the AI, that probabilistic intelligence alone does not run a company and deterministic systems do not reason, and that 'This is how every business will run.'
Follow any of these and your For You feed starts watching them — no settings page required.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Announcement-grade only
The deal, the 37-skill plugin, the AIforce/MCP harness and the availability window are all documented, but the substantive capability and governance claims trace to a single company press release that CNBC largely restates. There is no independent test, benchmark, customer account or audit detail for the routed-enforcement and central-permissions claims.
Pilot-stage, with prior Claude footprint
Claudeforce itself is limited to a select set of pilot customers with preview promised next month, so real usage is minimal. The score is not lower because Salesforce separately states Claude is already embedded across Agentforce and Slack, giving the partnership an existing deployment base independent of the new plugin.
Claims run ahead of a pilot
Language including 'AI CRO', 'a new era of enterprise AI', 'the UI is the AI' and 'This is how every business will run' is attached to a product that only pilot customers can use, with no published evaluation of the 37 skills. The gap is positive but not extreme because the underlying integration facts, the existing Claude footprint and the timeline are stated plainly.
Both parties selling the same story
One of two sources is Salesforce's own investor-relations release, published the same day as an earnings beat, while the stock sits well below its start-of-year level amid investor questions about AI vendors displacing application software. Anthropic gains distribution into CRM workflows and the companies describe themselves as strategic customers of each other, so every principal benefits from the announcement reading as complementarity rather than substitution.
Facts firm, effects unproven
Confidence is high on what was announced, when, by whom and on the market and revenue figures, all of which are consistently reported. It is low on whether the enforcement model, permissions design and seller productivity claims hold, since the cluster contains no independent evaluation and only two sources, one of them the vendor.
product
Slack Code makes coding agents taggable teammates. Your merge-approval policy is now overdue4 distinct publishers
invest
Salesforce's 14% cRPO is the best agent-demand read yet. Organic growth is 6.5%2 distinct publishers
invest
Salesforce bought $27.1B of its own stock in a quarter it grew 13%2 distinct publishers
invest
Salesforce's double digits, minus Informatica: agentic AI is real and still 2% of revenue1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.