What is Claudeforce?

What Is Claudeforce? A Complete Guide for Salesforce Teams

Claudeforce is the partnership between Salesforce and Anthropic, announced on August 26, 2026. It isn't a single product — it's the brand for a two-way integration: Anthropic's Claude becomes a core reasoning model inside Salesforce (including Agentforce and Slack), and Salesforce becomes usable from inside Claude through a plugin called "Salesforce in Claude," which ships with 37 prebuilt sales skills for things like meeting prep, deal health reviews, and pipeline updates.

The announcement moved fast from press release to practice: the plugin is in pilot with select customers now, open beta is planned for September 2026, and more skill packs — including revenue and quoting workflows — are slated for late 2026. Expect Dreamforce (September 15–17) to add detail; this guide is current as of September 9, 2026, and we'll update it as the beta ships.

This guide covers what Claudeforce actually is versus what people assume it is, what shipped and what's still roadmap, how the architecture works, the pricing picture, and what it means for your revenue data — including how a Salesforce-native platform like Kugamon fits into an AI-driven CRM workflow.

Claudeforce vs. Agentforce vs. "Salesforce in Claude" — Which Is Which

The naming causes most of the confusion, so here's the map:

The shorthand: Claudeforce is the umbrella; "Salesforce in Claude" is the seller-productivity piece; Agentforce remains the agent platform, now with Claude as a first-class brain.

What "Salesforce in Claude" Actually Does

The plugin exposes your Salesforce org to Claude through prebuilt skills — structured workflows rather than free-form prompting. The launch set of 37 covers seller work like meeting preparation, deal health checks, pipeline reviews, daily briefings, close plans, and CRM hygiene updates. An admin connects the org once; individual users then work through their existing Salesforce permissions, and actions route through Salesforce so business rules stay enforced.

Two caveats worth stating plainly. First, availability: as of early September 2026 this is a pilot with open beta "planned" — if you're not in the pilot, you can't use it yet. Second, scope: the launch skills are sales-productivity skills. Salesforce's roadmap lists skill packs for Service, Marketing, Commerce, Field Service — and Revenue, described as quoting "with context and governed logic" — as coming later. Quote-to-cash inside Claude is a promise with a "coming soon" label, not a shipping feature.

How It Works Under the Hood

Salesforce describes the architecture in three layers: an enterprise harness it calls AIforce, a "Headless 360" layer that exposes data, workflows, business logic, and permissions over MCP (Model Context Protocol), APIs, and CLI, and the Claude models themselves — served through Amazon Bedrock inside the Salesforce Trust Boundary, with zero data retention on the Claude model family and additional enterprise safeguards phasing in through fall 2026.

If you've been connecting Claude to Salesforce yourself via MCP — plenty of admins have — the honest difference is packaging and governance: one admin connection for the whole team, co-engineered skills, and centrally enforced permissions, instead of per-user tokens and homegrown prompts.

What It Costs

No rate card has been published. Reporting on the launch describes consumption-based metering through Salesforce (tied to license edition) for the Salesforce side, with Claude inference contracted separately with Anthropic — meaning, at least for now, two vendors on the bill and no single line-item price. If cost predictability matters to you, that's the first question for your account executive, in writing.

Shipped vs. Coming Soon (as of September 9, 2026)

What Claudeforce Means for Your Revenue Data

Here's the part that matters beyond the headlines: every layer of this architecture — Headless 360, MCP, permission-aware skills — operates on data that lives in Salesforce objects. When Claude preps a renewal call or checks deal health, it can only reason over what it can reach: native records, governed by your org's permissions.

That has a direct implication for revenue stacks. Quotes, orders, subscriptions, and invoices that live in a standalone tool and sync summaries into Salesforce are, at best, partially visible to this entire generation of AI. Revenue data on native Salesforce objects is fully in scope — the same thesis we laid out in Why AI Agents Need Native Revenue Data, which Claudeforce's architecture now validates from the vendor side.

And because Salesforce's own Revenue skills are still "coming soon," there's a practical note for quote-to-cash teams: Kugamon's objects are native Salesforce data reachable over MCP today, and we publish an open toolkit at github.com/kugamon: kugamon-skills for the full quote-to-renewal lifecycle, salesforce-core-skills (16 general Salesforce skills that work with any Salesforce MCP server), and salesforce-mcp-auto-auth-chrome, a Salesforce MCP server for Claude Desktop — so Claude-driven quoting workflows on Kugamon data don't have to wait for a roadmap. (For how the products relate on the Salesforce side, see What Is Agentforce Revenue Management?)

Frequently Asked Questions

Q: What is Claudeforce in one sentence?

Claudeforce is the Salesforce–Anthropic partnership, announced August 26, 2026, that makes Claude a core model inside Salesforce and Slack, and makes Salesforce usable from inside Claude via a plugin with prebuilt skills.

Q: Is Claudeforce a product I can buy today?

Not as such. Claude inside Agentforce and Slack is live; the "Salesforce in Claude" plugin is in pilot with open beta planned for September 2026. Pricing hasn't been announced.

Q: What's the difference between Claudeforce and Agentforce?

Agentforce is Salesforce's agent platform, primarily for customer-facing and workflow agents. Claudeforce is the partnership that, among other things, makes Claude a default reasoning model for parts of Agentforce and adds a seller-facing Claude plugin. They coexist.

Q: How is this different from connecting Claude to Salesforce with MCP myself?

Functionally similar plumbing, different governance: one admin-level connection, centrally enforced permissions, and co-engineered skills instead of per-user tokens and DIY prompts. Teams already using Claude with Salesforce's MCP or CLI get a managed version of what they built by hand.

Q: What does Claudeforce cost?

Unannounced. Launch reporting describes Salesforce-side consumption metering plus a separate Claude inference contract with Anthropic — two bills, no published rate card. Get specifics from your AE in writing before budgeting.

Q: Does it work with custom objects and ISV apps?

The architecture reads Salesforce objects through your org's permissions, which is exactly why native data matters: managed-package objects (like Kugamon's quotes, orders, and subscriptions) are Salesforce records and sit inside that reach. Data living in external tools behind a sync is only as visible as the sync makes it. Confirm specifics for any app you rely on during the beta.

Q: Can Claude create quotes in Salesforce through Claudeforce?

Salesforce's Revenue skill pack for that is announced but not shipped ("late 2026"). Quoting workflows on native revenue objects are possible today via MCP and published skills — that's the approach Kugamon supports with its open GitHub toolkit (kugamon-skills, salesforce-core-skills, and a Salesforce MCP server for Claude Desktop).

Q: Where can I read the official announcement?

Salesforce's press release of August 26, 2026 and the official page at salesforce.com/claudeforce, which includes the partnership FAQ.

Next Steps

If you're evaluating what an AI-driven CRM means for quote-to-cash, start with the architecture question — where does your revenue data live? Read Why AI Agents Need Native Revenue Data and What Is Quote-to-Cash?, explore Kugamon Quote-to-Cash, or schedule a demo and bring your Claudeforce questions. Primary sources: the official announcement and salesforce.com/claudeforce. No pitch — just honest guidance on a two-week-old product category.