The Platform Comes to You

The Platform Comes to You: How We’re Bringing the Full Power of Salesforce to Any Interface

Think about the last time you needed a specific piece of customer information. The status of a deal, the history behind an account, the open cases for a client you're about to call. Chances are you didn't go looking for it where it actually lives. You pinged a colleague, searched your inbox, or opened yet another tab.

That gap is what we kept coming back to. A company's knowledge is already in Salesforce. But for most people it sits behind a login, a navigation menu and a set of fixed screens. The data is there. Getting to it takes effort, and that's usually where good intentions about working with data quietly end.

We don't think people should have to adapt to the system. The system should show up where people work. That is the idea behind AIforce.

From destination to context

For a long time, using Salesforce meant going to Salesforce. One application, one interface. It worked, but it also set limits: on who could use the information, on how much of it they could see at once, and on how much value they could get out of it.

With AIforce, we bring Salesforce data, workflows, business logic and controls into the AI interfaces where people and agents already work. That includes tools like Claude and Slack, and of course Salesforce itself.

To be clear about what we mean: this is more than a CRM record shown in another window. You can ask a question and get an answer grounded in your own Salesforce data. You can update a record or start a workflow from wherever you happen to be. And agents can read across hundreds of records, draw on connected systems and look at the business as a whole, instead of one record at a time. The goal is that people and agents act on what they find, rather than sorting through lists of records.

What this looks like on a Tuesday afternoon

Take a sales lead who wants to know which key accounts have gone quiet. In Slack, she asks. The answer pulls together the accounts, the related support cases and the Slack threads around them. She reassigns an owner, creates a follow-up task and drafts a win-back email, all in the same conversation. That used to mean several tools and a good part of an afternoon.

The same logic applies to the way interfaces themselves get built. Instead of relying only on fixed layouts and fields, people can describe what they need and have a live interface put together around the task, showing the Salesforce information and actions that matter right now. Engine, a company that handles more than 800,000 customer inquiries a year, puts it simply: employees describe what they need and get a working interface based on existing data, instead of waiting for someone to build a dashboard or a report.

For teams that want to go further, we built this to be open. MCPs, APIs, plug-ins and skills can be used to create or extend interfaces, agents and views. Our Headless Toolkit exposes parts of the platform for this kind of work, and AgentExchange gives Salesforce and our partners a place to share what they build.

One foundation, many interfaces

The interface can change. What sits behind it should not. Data 360 brings together data, metadata and memory to give every request unified context. Customer 360 supplies the logic, permissions and actions behind processes in sales, service, marketing and commerce. Agentforce provides ready-to-deploy agents, and the tools to build your own. AIforce is how this foundation reaches the tools people already use.

In practice, this shows up in a few places:

• In Claude, Claudeforce provides Salesforce through a prebuilt MCP server, which removes much of the manual setup, authentication work and skill mapping that integrations usually involve.
• In Slack, teams can bring live context from Salesforce, Slack and other tools into one interactive space, filter it, comment on it and act on it together. Users can also create accounts, log call notes and update records with a simple prompt.
• In Salesforce, Agentforce Coworker works in the Lightning interface and reasons across accounts, activity and history. It can also call the specialized agents our customers have already built. 100,000 users activated Coworker within its first 35 days. Fulton Bank went from zero to more than 20 use cases in production within weeks, supporting around 3,000 users.

Trust comes first

Putting business context into more places raises a fair question: who gets to see what? We think it is the right question, and we designed for it from the start.

Requests run on the permissions and business rules you already have, so people and agents can only see and do what those controls allow. Actions route back through Salesforce. There is no new permissions model, no migration and no custom integration work: administrators connect once, and teams can start using it. And with Zero Data Retention, business data is used to answer the question at hand and is not retained by the model provider.

The more interfaces you add, the more important it becomes that the rules stay in one place. In our view, that is what makes it possible to use AI in daily work without losing control of the data behind it.

A question to take home

Every company has a version of the same problem: valuable knowledge, and too many steps between that knowledge and the person who needs it. We think the answer is not another application. It is a platform whose context travels with you.

So here is a question we'd leave you with after the summit: If your colleagues didn't have to open Salesforce to use everything that's in it, what would they do differently on Monday morning?