Salesforce Headless 360 Platform Built for Agents
- Anuj Vaishampayan
- 18 minutes ago
- 7 min read
TDX 2026 just wrapped up and I’m still processing it.
Every year, Salesforce makes big announcements. Every year, the community gets excited. And every year, the real question isn’t what got announced, it’s what actually gets adopted in the field. That’s where things get interesting.
This year felt different. It wasn’t a feature drop. It was an architectural shift. And if you’re a Salesforce developer or architect, you need to pay attention to what just moved underneath you.
The Headline: Salesforce Headless 360
Salesforce announced Headless 360 with a tagline that sums the whole thing up: “One platform. Anywhere humans and agents work.”
In plain terms, everything on the platform is now an API, an MCP tool, or a CLI command. Your data, workflows, business logic, and compliance controls. All accessible without ever opening a browser.
Parker Harris stood on stage and asked: “Why should you ever log into Salesforce again?”
That’s the company that built the most successful browser-based enterprise UI telling you the browser is optional. Salesforce said it made the decision two and a half years ago to rebuild the entire platform for agents. Headless 360 is the result.
The Three Pillars
Salesforce framed Headless 360 around three commitments: Build together with coding agents. Scale Agentforce with trust. Deploy on any surface.
Build means your coding agent, Claude Code, Cursor, Agentforce Vibes, now gets complete, live access to your Salesforce org. 60+ new MCP tools, 30+ preconfigured coding skills. Apex, LWC, React, Agent Script (now open source).
Scale is the architect-critical one. Testing Center GA in May with custom scoring evals, A/B Testing API in pilot, and Session Trace via OpenTelemetry for piping agent traces into Splunk, Datadog, or whatever observability stack you already run.
Deploy is the Experience Layer. Build a component once, render it natively in Slack, Teams, Mobile, Voice, ChatGPT, Claude, WhatsApp. No more rebuilding the same thing three times for three surfaces.
The Architecture Underneath
Here’s how I think about the Headless 360 stack as an architect: four layers, clean separation of concerns, each layer with a clear purpose:
Salesforce laid out their own version of this at the keynote, which is worth seeing alongside mine, same idea, more detail on what sits inside each layer:
Four layers, top to bottom:
System of engagement: where humans interact. Slack at the center, open to any workspace.
System of agency: Agentforce customer and employee agents. Open to any agent via MCP and A2A. You’re not locked in.
System of work: Customer 360. Sales, Service, Marketing, Commerce, Revenue Management, Platform. All the Industry Clouds. Integrates with any app.
System of context: Data 360. CDP, Federation, MuleSoft, Informatica, Tableau. Zero copy. Real-time. Any data lake or warehouse.
Underneath all of it, a Trust Layer that works with OpenAI, Anthropic, Gemini, LLaMA, Mistral, or open source. You pick the model. Salesforce provides the governance.
The Part Most People Are Sleeping On
The headline is access. Agents can reach into Salesforce programmatically. But the real story isn’t speed. It’s context.
A coding agent connected to a raw database doesn’t know that a customer has an open escalation, a renewal due in 30 days, a breached SLA, and a relationship owner who plays golf with their CFO every Tuesday. That context took years to accumulate. It lives in Salesforce.
Headless 360 exposes all of it programmatically. When an agent operates within Salesforce, it inherits your approval chains, sharing rules, business logic, and edge-case handling. You don’t rebuild any of it. You extend it.
That’s the real play. Not “Salesforce without a UI.” It’s Salesforce becoming the context engine for every agent you build.
This is also the strongest answer to the “vibe code your own CRM” narrative. Sure, you could build a CRM from scratch. But where’s twenty years of accumulated business context going to come from? You’re not just replacing software, you’re replacing institutional knowledge.
Open To Any Agent
One of the strongest statements Salesforce made is that Headless 360 is open. Not “open API” open, any agent, any model, any coding tool open.
Dozens of MCP providers across all four system layers — mcp-provider-apex, mcp-provider-lwc-experts, mcp-provider-testing, mcp-provider-devops, soql_query, metadata-experts, data-cloud-queries. That’s granular. That’s real. That’s a working set of tools any MCP-capable agent can pick up today.
For architects, this matters. You don’t bet on one model provider. You bet on the platform. If Claude gets better, great. If GPT-6 ships, great. If compliance mandates a specific model, fine. The platform adapts.
Other announcements worth your time
Agent Script is open source. Full spec, grammar, parser, compiler on GitHub. Claude Code can already generate it natively.
Agentforce Vibes 2.0. Browser-based IDE with Claude Sonnet as the default model. Free in every Developer Edition org through May 2026.
AgentExchange. AppExchange, Slack Marketplace, and the Agentforce ecosystem unified into one marketplace. 10,000+ apps, 1,000+ agents and MCP servers, $50M Builders Fund.
Slackbot as MCP client. Orchestrates across Agentforce agents, Slack apps, and AppExchange apps from one conversation. AI agents on Slack up 300% since January.
Deterministic → Probabilistic. Joe Inzerillo’s point worth remembering: agents are probabilistic systems. Different outputs from the same input. That needs fundamentally different testing. Testing Center, A/B Testing API, and Session Trace OTel API are built for exactly that.
My Honest Architect Take:
Every big Salesforce announcement is exciting and full of promise. And what I’ve seen over the years is that the real story is never the keynote, it’s what happens six months later. How these capabilities are implemented, adapted, and governed across enterprises with different maturity levels, data quality, and team structures.
With traditional software, 80% of the effort is the build and 20% is refinement. With agents, it’s flipped. Getting something working is the easy part. Making it production-ready is the other 80%.
That resonates because the hardest part is never building something new. It’s making it work within the reality of what already exists, inherited complexity, managed packages, undocumented decisions, the politics of change management. Nobody demos that part.
Where this gets interesting for Agentforce Revenue Management
Most of my work sits in Salesforce Industries and Agentforce Revenue Management, the full quote-to-cash lifecycle, where complexity is the default. ARM natively covers Product Catalog Management, CPQ, Dynamic Revenue Orchestrator, Rating, and Billing & Invoicing.
Headless 360 doesn’t just make one of these capabilities accessible to agents. It makes all of them accessible, together. The real unlock isn’t “an agent that can do CPQ.” It’s an agent that can reason across the full lifecycle, from catalog to quote to order to rating to invoice, because all of it sits behind a consistent programmatic interface.
For architects working in quote-to-cash, that changes the shape of what’s possible. Configuration intelligence, cross-lifecycle account insights, observability into complex orchestrations, conversational answers to questions that used to require pulling data from multiple modules, the surface area for meaningful use cases is broad.
This is genuinely an interesting space to watch. The specifics of which patterns win, which ones don’t, and where the real value lands will become clear as teams start building on it post-GA. That’s the part I’m most curious about.
The Exciting Part Is Real. So Is the Enterprise Reality.
Platform-level shifts like this don’t come along often, and we should be excited. What I’ve learned from years of enterprise implementations is that the most successful rollouts pair that excitement with honest evaluation. Not skepticism, evaluation.
A few practical questions worth thinking through:
How does this fit the way we build today? Teams with mature DevOps, CI/CD, and testing practices will have a head start.
What’s the operational model? Clear ownership, observability, and governance make agent rollouts stick, the same fundamentals that make any production system successful.
What’s the realistic investment? Model usage, testing, observability, upskilling, and the architectural work to make existing implementations agent-ready. The total picture is usually bigger than the headline cost.
How will users experience the change? Moving from UI-driven to agent-driven workflows touches training, change management, and UAT. The people change is a first-class workstream.
Which use cases genuinely move the needle? Not every workflow benefits from an agent. Picking the right handful, real volume, real pain, clear success criteria, is what separates a flagship adoption from a pilot that never ships.
The enterprises that get the most out of Headless 360 will evaluate thoughtfully, pick the right use cases, and build with a clear picture of cost, effort, and time to value.
At Stratus Carta, we love innovation, and we love making customers successful even more. That’s why we lean into platform shifts like Headless 360 with both excitement and honest advisory judgment. Once it’s generally available, we’ll experiment across use cases spanning catalog, quoting, orchestration, rating, and billing, hardening the patterns and building reusable architectural blueprints that our customers can trust.
Our job isn’t to chase every shiny thing. It’s to tell our customers what will genuinely move the needle for them and help them get there confidently.
Because new tools don’t make customers successful. Good judgment does. And that’s the part we bring to every engagement.
Powerful tools in the hands of thoughtful architects - that’s where real customer value gets built.
About the Author
Anuj Vaishampayan is a Director, Technical Architect at Stratus Carta with over 16 years of experience in the software industry. A recognized expert in Salesforce Industries (SFI) and Revenue Cloud, Anuj has led more than 20 Communications Cloud implementations.
With 14 Salesforce certifications and a background as a Principal Architect for Industries Clouds at Salesforce, he specializes in architecting scalable, industry-specific solutions that turn complex lead-to-cash challenges into streamlined business value.
Reference: Salesforce TDX 2026 keynote (April 15–16, 2026, San Francisco).
Images courtesy of Salesforce TDX 2026 keynote slides.