Salesforce Headless 360 and Agentforce Vibes 2.0 Revealed at TDX 2026

Salesforce is releasing what it calls ‘Headless 360’ along with Agentforce Vibes 2.0 at TrailblazerDX today, as “everything on Salesforce” becomes an API, MCP tool, or CLI command – all usable by agents.

Salesforce Headless 360 delivers new MCP tools and coding skills that give coding agents full access to your platform; a new experience layer that renders rich, native interactions across surfaces from Slack to Voice; and tools that give you control over how agents behave in production, before launch, and after. Let’s take a look at what’s been announced. 

Salesforce Headless 360 and Agentforce Vibes 2.0

Salesforce says that Headless 360 means developers can build on Salesforce “any way you want”, with more than 60 new MCP tools and 30 preconfigured coding skills empowering coding agents with complete, live access to your entire platform.

This includes all of your data, workflows, and business logic, directly in the coding agents you might already use, such as Claude Code, Cursor, Codex, and Windsurf.

Salesforce says that ‘Agentforce Vibes 2.0’, a ramped version of its vibe-coding tool, brings that same capability natively inside Salesforce – bringing multi-model support, including Claude Sonnet and GPT-5, and an AI “development partner” that understands your business.

Tim Combridge, Technical Content Writer at SF Ben and a keen vibe-coder, said:“Since Agentforce Vibes came out just prior to Dreamforce ’25, this tool has changed the way many work with Salesforce metadata. Admins and developers alike could benefit from the workflow speed enhancements that it brought, but the desire for more has kept us all hungry.

“We knew that OpenAI’s GPT-5 was the engine behind ‘Pro Mode’, and learned that we may be able to bring our own model ‘soon’. It looks like ‘soon’ is finally here with Agentforce Vibes 2.0! Now, those of us who prefer Claude Sonnet or another model can connect it to do what we need to do. A huge, welcome change for many who had been looking forward to a more capable Agentforce Vibes.”

Meanwhile, the DevOps Center MCP brings programmatic access into your CI/CD pipeline, and Natural Language DevOps means you can describe what you want to deploy, and agents will execute. 

“The build loop that used to require context-switching across four different tools now happens inside one connected experience – cutting cycle times by as much as 40%,” Salesforce claims.

And Native React support means developers who prefer full control over the visual layer can build custom interfaces and experiences with any design language or interaction model.

Adones Guerra, Tech Lead, Grupo Globo, said: “As a dev team lead, I’ve seen firsthand how Agentforce Vibes is supporting our Salesforce development. It’s been especially valuable for speeding up day-to-day tasks like metadata updates and quick tweaks, reducing manual work and boosting efficiency. 

“We primarily use it for repetitive, time-consuming work, from config analysis and metadata adjustments to generating boilerplate code, tests, and handling small refactors or pre-deployment checks. Vibes acts as a strong assistant that streamlines our workflows, and we’re excited to see how it evolves and drives even greater impact over time.”

Agentforce Experience Layer

Salesforce is positioning Slack as the place where “conversation lives”. Custom AI agents on Slack have grown 300% since January, the company says, adding that Slackbot is the “front door to the Agentic Enterprise”.

The Agentforce Experience Layer is a new UI service that separates what an agent does from how it appears, so agents can deliver interactive components like flight status cards, rebooking workflows, decision tiles, and data layouts. 

Salesforce says they render natively inside Slack, and across Mobile, ChatGPT, Claude, Gemini, Teams, or any client that supports MCP apps. “Build once, render everywhere your people already work,” the company says. 

Agents are probabilistic, not deterministic. They don’t behave the same way every time, and can come across unanticipated scenarios and reason their way to what Salesforce describes as “unexpected outcomes”. This is not due to any bug that needs fixing necessarily, but behavior to observe, evaluate, and tune.

Salesforce says it has built a new set of tools to provide control across every stage of the agent lifecycle.
Testing Center surfaces logic gaps, policy violations, and inconsistent outputs before launch, but Custom Scoring Evals score whether something made the right decision.

“You define what good looks like for your use case,” as Salesforce puts it, giving an example of a customer service agent who declines an out-of-policy refund while clearly explaining alternatives. Responses are evaluated against that standard, not just on whether it is executed.

Agent Script provides control over the agent itself before it gets tested, while you define which parts of its behavior must follow explicit business logic and which parts can reason freely.

After launch, Observability and Session Tracing show why something happens, and when an agent drifts, you can find the cause in “hours, not weeks”, Salesforce says. A/B Testing allows you to run several agent versions against real traffic simultaneously, with data-driven decisions about what to promote to production.

And for enterprises running agents across multiple platforms and vendors, the new Agent Fabric brings them under one governed control plane, with deterministic orchestration and centralized agent, tool, and LLM governance across your entire AI landscape.

Why Does the Platform Still Matter?

Salesforce says that, even though agents can call any API, the platform still matters because intelligence alone is “just inference” and, in reality, needs context, workflows, trust, and the right engagement layer to actually do anything.

Coding agents connected to a raw database do not know that a customer has an open escalation, a renewal due in 30 days, a breached support SLA, and a relationship owner with a personal connection to their CFO – context that took years to accumulate, living in Salesforce. 

Now, Data 360 exposes it as an API, an MCP tool, a CLI command, so coding agents can reach it from anywhere, “without touching a UI”. 

Context is only part of what separates a coding agent that writes code from one that understands your business, according to Salesforce. Agents also need workflows. 

When an agent operates inside Salesforce, it inherits approval chains, business rules, and edge-case logic built years ago, and does not have to rediscover or approximate them. 

Then there is trust: permissions, sharing rules, compliance controls, which IT and security teams already approved. 

“In most platforms, you rebuild that trust layer from scratch every time you deploy something new. In Salesforce, it is already there. The guardrails your organization relies on are the same guardrails your agents operate within,” Salesforce says. 

“Most vendors can give you one layer. Some can give you two. Only Salesforce gives you all four, already integrated, already running inside your enterprise.”

Developers are building on infrastructure already running inside their enterprise, with the data, the workflows, the permissions, and the engagement layer already in place. 

Final Thoughts

Parker Harris recently asked, “why should you ever log into Salesforce again?”. A provocative question, perhaps, from the Co-Founder of Salesforce. In its move towards becoming (and enabling) the ‘agentic enterprise’, Salesforce made a decision to rebuild Salesforce for agents. 

“Instead of burying capabilities behind a UI, expose them so the entire platform will be programmable and accessible from anywhere,” Salesforce said as it announced Headless 360. Agents now orchestrate apps, workflows, and business logic, without having to rebuild any of it.