If you just got back from Dreamforce, there is a decent chance someone at your company has already asked you some version of:
“So what is AIforce, and how is that different from Agentforce and Claudeforce?”
Fair question.
Salesforce introduced enough new terminology this year that it can sound like these are three competing AI products. They aren’t.
The easiest way to think about it is this:
| Product Name | Plain English |
|---|---|
| AIforce | Makes Salesforce available to AI surfaces, wherever you want to work |
| Agentforce | Lets you build and run AI agents that actually do work |
| Claudeforce | Brings Salesforce directly into Claude |
The more important part is understanding how they fit together.
Start with AIforce
Historically, if you wanted something from Salesforce, you went into Salesforce.
You opened the CRM, found the account, opened the opportunity, looked at the activity history, ran the report, clicked through some related records, and eventually figured out what you needed to know.
AIforce changes that model.
AIforce is Salesforce’s new interface layer that makes the data, workflows, business logic, permissions, security, and context already inside Salesforce available to other AI experiences. That could be Claude, Slack, an Agentforce experience, or potentially another AI application your company builds or uses.
Instead of saying:
“Come into Salesforce to use Salesforce.”
The new model is:
“Bring Salesforce to wherever the user or AI is already working.”
And if the name AIforce feels particularly new, that is because it is. Salesforce officially renamed Headless 360 to AIforce on September 4, 2026. Salesforce’s own release notes say the name changed while the underlying functionality remained the same.
That is also why we would not think of AIforce as another application someone needs to go buy and implement. It is better understood as an architectural and interface layer across the Salesforce platform.
Then there’s Agentforce
Agentforce answers a different question.
If AIforce is about where AI can interact with Salesforce, Agentforce is about the AI agents doing the work.
Agentforce is Salesforce’s platform for building, deploying, governing, and running AI agents. These agents can reason over business context, decide what steps need to happen, interact with systems, and execute work within the guardrails you give them. Salesforce supports agents across areas including sales, service, commerce, and employee workflows.
Think about an agent that can:
Handle a customer service request. Qualify or research a prospect. Manage parts of a sales process. Schedule an appointment. Work a case. Or execute a multi-step business process without someone manually clicking through every step.
That’s Agentforce.
So AIforce and Agentforce aren’t replacements for one another. AIforce makes the Salesforce platform accessible to AI experiences, and Agentforce gives you Salesforce-native AI agents capable of doing jobs.
Where Claudeforce fits
This is probably where the naming gets the most confusing.
Claudeforce is the Salesforce + Anthropic partnership and product experience that brings Salesforce into Claude. Its first major product is Salesforce in Claude.
Instead of a salesperson opening Salesforce to research an account, they can stay inside Claude and ask something like:
“Give me a briefing on the Acme account, tell me what has changed recently, show me the biggest risks in the opportunity, and tell me what I should discuss in my meeting this afternoon.”
Claude can use live Salesforce context to produce the answer without the seller having to manually hunt through CRM records.
Salesforce in Claude launched with 37 prebuilt sales skills covering areas such as account research, meeting preparation, deal health, pipeline analysis, and prospecting. Salesforce announced during Dreamforce that the experience is now available in beta.
One important caveat for anyone running back to their sales team and promising full CRM automation tomorrow: Salesforce’s current release notes describe the initial beta experience as read-only, meaning Claude can surface and reason over Salesforce information but cannot yet create, update, or delete Salesforce records through that beta experience. The broader Claudeforce direction includes governed actions, but there is a difference between the long-term vision and what the current beta supports today.
…But you can still use Salesforce MCP with Claude
For those of us who have already been using Claude with Salesforce, you’re already familiar with the ability to query Salesforce data from Claude, and build artifacts, documents, and interfaces that interact with the data. This doesn’t supplant any of these capabilities, but it does introduce a new productized feature set that makes this a lot easier for companies who want it to be more plug and play.
The easiest way to understand all three
Imagine a salesperson is working in Claude before an important customer meeting.
They ask Claude to analyze the customer, the opportunity, recent activity, pipeline history, and any potential risks.
Claudeforce is the experience that lets the seller work with Salesforce from inside Claude.
AIforce is the architecture that allows an external AI experience like Claude to securely access Salesforce’s trusted business context.
Agentforce is the agent platform you can use to create AI workers that independently execute business processes and tasks.
Three different layers. Three different jobs.
The real story told at Dreamforce
The interesting part is the shift in how Salesforce thinks people will use enterprise software.
For the last 25 years, SaaS companies competed over who could build the best application interface. Users logged into the application, navigated the screens, clicked the buttons, searched for information, and executed workflows inside the software. AI is starting to break that model.
Users increasingly want to tell an AI what they need and have the AI figure out which data, applications, workflows, and actions are required to get there.
AIforce effectively says the value of Salesforce does not have to be the Salesforce UI. The real value is the enormous amount of customer data, metadata, permissions, workflows, business logic, and organizational knowledge sitting underneath it.
Agentforce lets AI agents work against that foundation, and Claudeforce equips Claude to become one of the interfaces with pre-built skills and capabilities that people use to access it.
And Salesforce says AIforce is launching not only with Claudeforce, but also Slackforce and Agentforce Coworker, which makes the strategy even clearer: Salesforce wants its platform to power work regardless of which AI interface sits in front of it. citeturn893526search0
What to tell the business
Don’t come back from Dreamforce and tell your executive team that Salesforce just launched three more AI products you need to evaluate.
Tell them this:
Salesforce is becoming less dependent on Salesforce being the place where users actually work. AIforce opens Salesforce’s data and business logic to AI interfaces. Agentforce gives companies AI agents that can perform work against that environment. And Claudeforce makes Claude one of the places employees can interact with Salesforce.
The interface is becoming interchangeable. The data, context, business processes, permissions, and ability to take action are becoming the important part.
And that is a much bigger change than another “Force” product name.