For years, using Salesforce has meant opening the platform, finding the right object, navigating through screens, checking records, and then taking action. That model is now changing. At Dreamforce ’26, Salesforce introduced AIforce, a new live interface layer designed to bring Salesforce data, business logic, workflows, security, and AI capabilities into the places where people already work. Instead of always asking users to come to Salesforce, Salesforce is starting to bring its capabilities to the user.
The bigger idea is simple: what if the interface is no longer a fixed set of screens, buttons, and fields, but something that can be created dynamically through AI? That is the thinking behind AIforce.
The “SaaS Apocalypse” Is Not Really About the End of Software
There has been a lot of discussion around the so-called “SaaS apocalypse” and whether AI could eventually replace traditional software. The Dreamforce keynote offered a different way to look at it. The argument is not that software is disappearing. Instead, software may be changing from applications that require humans to perform every step into systems where AI and humans work together.
Large AI models are powerful, but a model by itself does not know the exact truth about how a particular company operates. An AI model can reason and generate answers. But an enterprise also needs reliable data, business rules, workflows, permissions, security policies, and processes.
That creates an important distinction:
AI provides probabilistic intelligence. Salesforce provides deterministic enterprise context.
The real opportunity comes from bringing the two together.
Salesforce has spent years building the data, application, semantic, security, and workflow layers that companies depend on. AIforce is designed to make that foundation available through AI-driven interfaces.
From DOS to AI: Another Interface Revolution
The way people interact with technology has changed several times. We moved from:
DOS → Graphical Interfaces → Web → Mobile → AI interfaces
Each transition changed how people interacted with software. The next change could be just as significant. Traditional software gives users a predefined interface. You learn where things are, which buttons to click, which fields to fill in, and which screens to navigate. AI interfaces can work differently.
Instead of navigating through a fixed experience, users can describe what they need. For example, instead of opening several Salesforce reports and dashboards, a sales leader could ask:
“Show me our 10 biggest deals this week, which ones are at risk, and the common reasons behind those risks.”
The AI can then reason across the available Salesforce context and present the relevant information. Salesforce describes this new generation of interfaces as dynamic, intelligent, composable, and live.
So, What Exactly Is AIforce?
AIforce is designed as a live interface layer for Salesforce. Its purpose is to make Salesforce capabilities available across different AI and work environments rather than limiting interaction to the traditional Salesforce UI. Salesforce says AIforce can bring together:
- Salesforce data
- Business logic
- Workflows
- Business semantics
- Permissions
- Security
- Governance
- AI agents
This means the AI interface is not supposed to operate separately from Salesforce’s underlying controls. Instead, it works on top of the existing Salesforce foundation. That distinction is important. AIforce is not simply another chatbot sitting beside Salesforce.
The goal is to allow AI to reason and take action using the same trusted enterprise foundation that already powers the business.
AIforce Is About More Than Asking Questions
One of the interesting ideas from the keynote is that AIforce is not limited to answering questions. The interface can potentially help users with three broad areas:
1. Using applications
Users can ask AI to retrieve information, analyze data, summarize business situations, and perform actions.
2. Building applications
AI can help create interfaces, views, and experiences based on what the user describes. Instead of starting with a blank screen and deciding which components to place on it, the user can describe the experience they want.
3. Operating applications
AI can also become part of the ongoing administration and operation of business applications. This creates a much more conversational way of interacting with enterprise software. The Headless Toolkit provides the underlying building blocks for this approach, including APIs, MCPs, plug-ins, skills, and developer tools.
Salesforce Is Separating the Platform From the UI
This may be one of the most important ideas behind AIforce. For years, people often thought of Salesforce as the application they opened in a browser. But Salesforce’s platform contains much more than the screens users see. Behind the UI are:
- Customer data
- Relationships between records
- Automation
- Business rules
- Permissions
- Security
- Workflows
- Metadata
- Integrations
- Business processes
The Dreamforce message is that the UI does not have to be the only way to access those capabilities. The platform can become the foundation, while different AI interfaces sit on top of it. Salesforce’s Headless Toolkit is intended to expose those capabilities through technologies such as APIs and MCP, allowing Salesforce functionality to be used in different environments.
This is why AIforce is better understood as an evolution of Salesforce’s platform rather than simply another Salesforce feature.
AIforce Launches With Three Major Experiences
Salesforce is launching AIforce with three major experiences: Claudeforce, Slackforce, and Agentforce Coworker.
Claudeforce: Salesforce Inside Claude
Claudeforce brings Salesforce capabilities directly into Claude. Salesforce says Salesforce in Claude combines a prebuilt Salesforce MCP server with 37 ready-to-use sales skills, covering activities such as prospecting and pipeline work. This changes the normal workflow. Instead of:
Open Salesforce → Find the opportunity → Review the information → Analyze it → Take action
the interaction can become: Ask Claude → Claude uses Salesforce context → Get the answer or take action
The Dreamforce demonstration showed this idea through Claude/Cowork, where Salesforce information could be used to understand pipeline, opportunities, deals, and broader business trends. For example, a sales leader could ask questions such as:
- What are our biggest deals this week?
- Which deals are most at risk?
- What themes are appearing across these opportunities?
- What can we do to improve our chances of winning them?
The important part is that the AI is working with the company’s Salesforce context rather than operating only from general model knowledge.
Slackforce: Salesforce Where Teams Already Work
The second experience is Slackforce. Slackforce brings Salesforce context into conversations and workflows in Slack. Salesforce describes Slackforce as a live interface that allows teams to interact with Salesforce through conversations and interactive Slack Surfaces. This can reduce the need to switch between applications.
Imagine a team discussing a customer in Slack. Instead of someone saying: “I’ll open Salesforce and check.” The Salesforce information can become available directly within the conversation. Teams can work with customer information, update records, and interact with Salesforce without constantly leaving the environment where the discussion is happening.
Agentforce Coworker: AI Inside Salesforce
Not everyone will want to leave the Salesforce interface. That is where Agentforce Coworker fits. Agentforce Coworker brings an AI teammate into the Lightning experience and works within Salesforce’s existing permissions and business rules. So the future is not necessarily:
Salesforce UI disappears completely.
Instead, Salesforce is creating multiple ways to interact with the same underlying platform. You might use:
- Salesforce with Agentforce Coworker
- Claude with Claudeforce
- Slack with Slackforce
- Other AI interfaces powered by the same Salesforce foundation
The Most Important Part: Trust
One of the strongest ideas from the keynote is that Salesforce does not see the UI as the main product. The deeper value is the trusted business foundation underneath it. A company has spent years configuring:
- Who can access certain records
- Which users can perform specific actions
- What approvals are required
- Which business rules must be followed
- How workflows operate
- How customer data is protected
AI needs to operate within those boundaries. That is why AIforce is designed to use existing Salesforce permissions and business rules. Salesforce says every request runs through those controls, so the agent sees and acts according to what the interacting user is allowed to access. This is especially important in enterprise environments.
A general-purpose AI model may be excellent at reasoning, but enterprise AI also needs to understand what it is allowed to do.
Probabilistic AI Meets Deterministic Systems
This is another major theme from the Dreamforce keynote. Large language models are probabilistic. They generate responses based on patterns and reasoning. Enterprise systems such as Salesforce are different. They contain structured information and defined rules.
For example:
AI: “Based on the available information, this opportunity appears likely to close.”
Salesforce: “This opportunity is $500,000, belongs to this account, is currently in this stage, and this user has permission to access it.”
The combination is powerful. AI can provide reasoning and synthesis. Salesforce can provide the trusted business context. Together, they can create an enterprise experience that is both intelligent and grounded. This idea also appears in Salesforce’s description of AIforce as bringing data, workflows, business logic, semantics, permissions, security, and governance into AI interfaces.
Why “AI Replaces the UI” Can Be Misleading
The phrase “AI replaces the UI” sounds dramatic. But it does not necessarily mean that Salesforce screens, Lightning, or applications suddenly disappear. The more accurate interpretation is that the traditional UI no longer has to be the only interface. Salesforce itself continues to provide the underlying applications, data, workflows, and platform.
AI becomes another way to interact with them. In some situations, a traditional screen may still be the best option. In others, a conversational request may be faster. And in some cases, an AI-generated interface could be created specifically for the task at hand. That is the bigger shift.
What Does This Mean for Salesforce Admins?
For Salesforce Admins, AIforce could change the type of work they focus on. Admins have traditionally spent significant time configuring:
- Page layouts
- Fields
- Record types
- Flows
- Permissions
- Reports
- Dashboards
- User experiences
Those skills are not suddenly irrelevant. In fact, the underlying configuration may become even more important because AI needs reliable business context to work effectively. However, the role could gradually move toward designing the business foundation that AI interacts with. Admins may increasingly think about:
- What data should AI access?
- What actions should AI be allowed to perform?
- Which business rules must always be enforced?
- How should users interact with agents?
- How can we make our Salesforce data trustworthy enough for AI?
This makes Salesforce administration less about simply configuring screens and more about creating a reliable platform for both humans and agents.
What Does This Mean for Salesforce Developers?
Developers also have an important role in this new architecture. The traditional approach has often been to build custom components, applications, integrations, and APIs. The Headless Toolkit expands that model by exposing Salesforce capabilities through APIs, MCPs, skills, plug-ins, and other developer tools. (Salesforce). This means developers can increasingly build experiences where Salesforce is the backend foundation but the user interaction happens somewhere else.
The development question may shift from: “How do I build this Salesforce screen?”
to: “How do I expose this Salesforce capability so humans and agents can use it?”
That is a significant change in mindset.
The Siemens Example: AI in the Real Enterprise
The Dreamforce keynote also highlighted Siemens as an example of how enterprise AI can work with real operational data. The example is particularly interesting because industrial environments cannot simply depend on AI making unsupported guesses. The transcript emphasizes that Salesforce serves as a backbone for information about products, parts, manufacturing instructions, and service.
The goal is to bring that trusted information together with AI so sales and service teams can work faster while remaining connected to the company’s operational knowledge. This illustrates the broader idea:
Enterprise AI needs more than intelligence. It needs context.
Zero Data Retention and Governance
Enterprise adoption also depends on trust and governance. Salesforce says AIforce operates with existing permissions and business rules and supports Zero Data Retention, meaning business data used to answer a request is not retained by the model provider.
These controls matter because AI is moving from simply generating text to accessing information and potentially taking actions. The more powerful AI becomes, the more important governance becomes.
Companies need to know:
- What data can AI access?
- Who can use it?
- What actions can an agent perform?
- Which rules apply?
- Where does the data go?
- What happens when an AI request triggers an action?
AIforce is designed around the idea that these controls should travel with the Salesforce platform.
The Headless Toolkit Is the Foundation
AIforce does not appear out of nowhere. It builds on Salesforce’s broader move toward a more headless architecture. The Headless Toolkit exposes Salesforce capabilities through technologies such as:
- APIs
- MCPs
- CLI tools
- Plug-ins
- Skills
- Developer tools
Salesforce describes this as making its capabilities accessible wherever humans and agents work.
Is Salesforce Going Away?
No. The more interesting possibility is that the way people use Salesforce is changing. The Salesforce platform remains responsible for the things enterprises depend on: Data + Logic + Workflows + Security + Governance + Business Context.
AI becomes a new interaction layer on top. So instead of thinking:Salesforce vs. AI
it may be more useful to think:
Salesforce + AI
Salesforce provides the trusted foundation. AI provides reasoning and a more flexible way to interact with that foundation.
What Salesforce Professionals Should Watch Next
AIforce could have a broad impact across the Salesforce ecosystem. Admins should pay attention to data quality, security, permissions, automation, and governance. Developers should understand APIs, MCP, agent skills, integrations, and headless architecture. Architects should think about how data, AI, applications, and governance fit together.
Business users should understand how to communicate requirements to AI clearly. And everyone working in the ecosystem should understand one important change: The Salesforce UI may no longer be the starting point for every Salesforce task. The platform can increasingly meet users wherever they work.
Final Thoughts
AIforce is not simply another AI feature added to Salesforce. It represents a different way of thinking about enterprise software. For decades, people learned how to use applications by learning their interfaces. Now AI creates the possibility of making the interface more flexible and task-focused.
The Salesforce platform remains the foundation: the data, business logic, workflows, permissions, security, and governance that companies have built over many years. AI adds another layer: reasoning, conversation, synthesis, and action. The result is a future where users may not always ask:
“Where do I click?”
Instead, they may simply ask: “What do I need to get done?”
And Salesforce is building toward a world where the interface can appear around that answer. That is what makes AIforce one of the more interesting developments to watch from Dreamforce ’26.
