Salesforce announced a sweeping platform overhaul at TDX 2026, its annual developer conference in San Francisco. The initiative, called Headless 360, exposes every capability in Salesforce as an API, MCP tool, or CLI command so that AI agents can operate the entire system without a human ever opening a browser.

It ships more than 100 new tools and skills available to developers immediately.

“We made a decision two and a half years ago: Rebuild Salesforce for agents,” the company said. Jayesh Govindarajan, EVP of Salesforce and one of the key architects of the initiative, described it as “rooted not in marketing theory but in hard-won lessons from deploying agents with thousands of enterprise customers.”

What Headless 360 Actually Does

The announcement organises around three pillars.

Build any way you want delivers more than 60 new MCP tools and 30-plus preconfigured coding skills. External coding agents (including Claude Code, Cursor, Codex, and Windsurf) get complete live access to a customer’s entire Salesforce org, including data, workflows, and business logic. Developers can build agent automations without context-switching between tools.

Deploy on any surface introduces the Agentforce Experience Layer. Agents can now function across Slack, Microsoft Teams, and ChatGPT without requiring custom code for each surface. The same agent logic runs everywhere.

Build agents you can trust at scale adds lifecycle management tools covering testing, evaluation, and orchestration. Agentforce Vibes 2.0 and Session Tracing are generally available now. The Testing Center and Salesforce Catalog roll out in May and June.

Why This Matters Beyond the Product Announcement

Salesforce calling its own platform “infrastructure for AI agents” is a significant shift in how enterprise software vendors are positioning themselves.

For years, CRM was a system of record, a place where customer data lived. Headless 360 reframes it as a system of execution. The data still lives there, but now the agents are doing the work, not just storing the output of human work.

That distinction matters for how businesses think about their software stack. The question is no longer just “what data does this tool hold?” but “what can an agent do inside this tool without anyone being in the loop?”

For Salesforce customers (roughly 150,000 organisations use the platform), the practical implication is that existing data, workflows, and business logic can now be accessed programmatically by AI agents. You don’t need to rebuild your CRM to get agent-native behaviour. You get it through MCP.

The MCP Moment

The Headless 360 announcement is also notable for how heavily it leans into the Model Context Protocol standard. Sixty new MCP tools is not a small bet. It signals that Salesforce sees MCP as the emerging connective tissue between AI agents and enterprise software, and is building toward it seriously.

This aligns with a broader pattern across enterprise vendors in 2026. The companies investing in MCP-native interfaces are the ones positioning for a future where AI agents are the primary users of business software, not the people working within it.

What This Means for Business

For business owners and operations teams, the Headless 360 launch raises a practical question: if AI agents can now run your CRM end-to-end without human input, what does your workflow actually look like in 12 months?

A few things to think through:

What’s automatable today. Salesforce already had significant automation capability. Headless 360 makes it agent-accessible. If you’re running sales workflows, support queues, or marketing sequences inside Salesforce, the building blocks to automate them with agents are now there.

The governance gap. Salesforce is shipping testing and evaluation tools alongside the automation capabilities, which is the right move. Any organisation deploying agents inside their CRM needs to know how to monitor agent behaviour, catch errors, and maintain audit trails. This is not optional.

The vendor lock-in question. Deeper agent integration with any platform is valuable, and also a tighter dependency. That’s not a reason to avoid it, but it’s worth being deliberate about which platforms you’re building your agent workflows on.

For a deeper walkthrough of tools like this and how they fit together, the free Working With Claude field guide covers the ecosystem end to end. Get the guide.