Why Salesforce Now Has Four AI Names
Salesforce's AI portfolio has layered up over a decade. Einstein arrived in 2016 as predictive intelligence embedded in the clouds. Agentforce arrived in 2024 as the autonomous agent platform — and absorbed the Einstein Copilot name in early 2025. Slack, acquired in 2021, is now being positioned as the conversational work surface of the agentic enterprise, with Surfaces as its AI-generated workspace format. And in August 2026, Salesforce and Anthropic announced Claudeforce, bringing Claude's reasoning to Salesforce data and vice versa.
The common thread is Salesforce's stated direction: software is shifting from a fixed interface you navigate to a system that powers every interface — the CRM app, the agent, the Slack channel, or Claude. Each of the four names owns a different layer of that vision.
The Four Tools, Side by Side
What each one is, how Salesforce positions it, and where it fits best.
Einstein
The predictive and embedded AI layer
Salesforce's original AI brand, embedded across the clouds since 2016. Einstein scores leads and opportunities, forecasts, classifies cases, recommends next best actions and powers embedded generative features such as email drafting and summaries.
Salesforce's positioning: Einstein is the always-on intelligence inside the CRM you already own. It works on your historical Salesforce data and is largely bundled into editions or sold as per-user add-ons.
Best for: Organisations that want better prioritisation, forecasting and productivity inside Sales, Service and Marketing without changing how teams work.
Agentforce
The autonomous agent platform
Salesforce's agentic layer: digital workers that reason, plan and execute multi-step work through the Atlas Reasoning Engine, topics, actions and guardrails — across service, sales, field service, HR, IT and more, on web, phone, Slack and apps.
Salesforce's positioning: Agentforce is where Salesforce's platform investment is concentrated. It is consumption-priced, runs on Data 360 grounding, and is positioned as a digital labour platform rather than a feature.
Best for: Organisations ready to automate bounded, high-volume workflows end to end — case resolution, lead qualification, scheduling — with governance and human escalation built in.
Slack Surfaces
The collaborative work surface
AI-generated, interactive workspaces created inside Slack from natural-language prompts — dashboards, decks, reports, polls, calculators and lightweight microsites — grounded in Salesforce and other Slack-connected data, updated live and shared in channels.
Salesforce's positioning: Slack is being positioned as the conversational operating system for the agentic enterprise: the place where humans and agents decide together, with Salesforce data surfaced in the flow of work instead of a separate app.
Best for: Teams that live in Slack and want live pipeline reviews, service huddles, campaign tracking and decision-making without switching into the CRM interface.
Claudeforce
The Salesforce + Anthropic partnership
An expanded Salesforce–Anthropic partnership that brings Claude's reasoning together with Salesforce data, workflows, business logic and governance. It launches with Salesforce in Claude — a plugin with 37 prebuilt sales skills — plus Claude inside Agentforce and Claude as the default model in Slack experiences.
Salesforce's positioning: Claudeforce extends Salesforce beyond its own UI: through the AIforce harness (MCP servers, APIs, CLI tools), Claude can reason over live revenue context and take governed action where sellers and knowledge workers already work.
Best for: Organisations whose teams already use Claude and want trusted, permission-aware access to Salesforce data and actions inside that experience — currently pilot, with open beta planned.
Salesforce's Strategy: One Platform, Every Interface
Einstein keeps the CRM intelligent
Predictive scoring, forecasting and recommendations keep the core clouds valuable and sticky, and they run on the data customers already have.
Agentforce monetises digital labour
Consumption-priced autonomous agents are Salesforce's growth engine — a new revenue line that scales with work performed, not seats.
Slack becomes the front door
Surfaces, Slackbot and Claude in Slack turn the channel into the place where humans and agents decide and act together — reducing dependence on the classic CRM UI.
Claudeforce meets users where they already are
Through the AIforce harness, Salesforce data, workflows and governance travel into Claude and other interfaces — so Salesforce remains the system of record even when it is not the screen.
How to Choose: KVP's Decision Guide
When: adopt Einstein any time — it is low-risk and embedded. Pilot Agentforce when you have a bounded, high-volume workflow with clean data and a clear escalation path. Introduce Slack Surfaces when Slack is already your collaboration hub. Evaluate Claudeforce as it moves from pilot to open beta if your teams already work in Claude.
Why not everything at once: each layer multiplies the data-quality, permission and governance demands of the one below it. An Agentforce agent grounded in bad data fails loudly; a Slack Surface or Claude skill grounded in bad data fails quietly, in front of more people.
KVP View: What We See in Real Implementations
Einstein is the most under-used asset customers already own. Across our Salesforce implementation and support engagements, we regularly find orgs paying for predictive capabilities that were never switched on or never tuned. Turning on lead and opportunity scoring is often the fastest AI win available.
Agentforce succeeds when the use case is narrow and the data is clean. In our agentic AI work, the pilots that reach production start with one bounded job — a service triage agent, a lead qualifier — with explicit topics, actions and guardrails. Broad "do everything" agents stall in testing.
Adoption is decided in the flow of work, not in the CRM. Our experience with Slack integrations and collaboration-led rollouts is consistent: capabilities that meet users in the tools they already open every morning get used; capabilities that require a new habit get ignored. That is the real promise of both Slack Surfaces and Claudeforce.
Governance is the multiplier. Every layer inherits your permission model. Customers who invest in data quality, least-privilege access and monitoring early can adopt each new layer in weeks; customers who skip that work re-do it under pressure later.
Five Steps to Get Started
Audit what you already own
Inventory the Einstein features, Agentforce entitlements and Slack capabilities in your current licences before buying anything new.
Fix the data foundation
Scoring, agents, surfaces and Claude skills are all only as good as the CRM data and permissions underneath them.
Pick one workflow per layer
One Einstein model, one Agentforce agent, one Slack Surface — prove value narrowly before scaling.
Design governance once, reuse it everywhere
Least privilege, monitoring and human escalation patterns should carry across every layer you adopt.
Review the roadmap quarterly
Claudeforce is in pilot with open beta planned, and packaging across all four is evolving — confirm current availability with Salesforce before committing.
Official Sources
Availability, packaging and pricing for Agentforce, Slack Surfaces and Claudeforce are evolving; confirm current terms with your Salesforce account team.