Salesforce on Wednesday unveiled Headless 360 at its annual TDX developer conference in San Francisco, a new layer that lets AI agents work across the company’s platform without opening a browser. The system exposes every major capability in Salesforce as an API, an MCP tool or a CLI command, and the company said it shipped more than 100 new tools and skills that developers can use immediately.
The pitch is straightforward: Salesforce wants its platform to be usable by agents as well as people. Rather than putting features behind a screen, the company is making them programmable from anywhere, with more than 60 new MCP tools and 30-plus preconfigured coding skills included in Headless 360.
That effort did not begin this week. Salesforce said it made the decision two and a half years ago to rebuild Salesforce for agents, and the launch lands in a market that has grown wary of what AI could mean for traditional software subscriptions. The iShares Expanded Tech-Software Sector ETF has fallen roughly 28% from its September peak as investors have sold off software names across the sector.
The pressure is not abstract for Salesforce. The company framed Headless 360 as a response to the idea that AI could make older SaaS models obsolete, and it is betting that the answer is to let outside coding tools plug directly into a customer’s Salesforce org. Claude Code, Cursor, Codex and Windsurf can now get live access to data, workflows and business logic, turning the platform into something that can be operated by agents rather than just clicked through by employees.
Salesforce also used the conference to show where the product is headed next. Agentforce Vibes 2.0 includes an open agent harness that supports both the Anthropic agent SDK and the OpenAI agents SDK, along with multi-model support for Claude Sonnet and GPT-5. In a keynote demo, developers built a fully functional partner service application using React instead of Lightning, underscoring the company’s push to make the system more flexible for outside builders.
Jayesh Govindarjan, the company executive helping lead the effort, said the harder problem is no longer just building an agent at all. The challenge, he said, is creating the full lifecycle of an agentic system for customers using any stack, whether it is Salesforce’s or someone else’s. That is the bet behind Headless 360: if agents are going to become the interface, Salesforce wants its platform to be the plumbing they run on.
For crm stock, the question is not whether Salesforce is adding AI features; it is whether it can turn them into a reason customers stay inside its ecosystem. Wednesday’s launch answers that directly. Salesforce is not treating AI as a layer on top of its software. It is rebuilding the software so AI can use it first.
