Salesforce Adds MCP Support to Headless 360 for AI Agents

What happened

Salesforce has added support for Model Context Protocol (MCP) to Headless 360, its infrastructure layer for building AI agents. The update allows AI agents to discover and act on existing enterprise workflows and systems directly, removing the need for teams to build bespoke point-to-point integrations for every application an agent needs to reach.

In practice, this means an agent built on Headless 360 can query and trigger processes across connected systems using a shared protocol, rather than relying on custom-coded connectors maintained separately for each data source or application.

Why it matters

The move reflects a broader shift in enterprise AI architecture: away from agents that are hard-wired to specific systems, and towards agents that can flexibly discover and use whatever tools and data sources they are permitted to access. For organisations running large, fragmented technology estates, this lowers the engineering overhead of scaling agentic AI across departments, since new systems can, in principle, be exposed to agents without a fresh round of custom integration work each time.

For leaders overseeing digital transformation and AI adoption, the significance lies less in Salesforce's product roadmap and more in what it signals about where the industry is heading: standardised protocols like MCP are becoming the connective tissue that lets AI agents operate across a company's existing software estate, rather than requiring that estate to be rebuilt around the AI.

The Renascence take

Protocol-level interoperability is an operations story before it is a customer-experience story — but its downstream effects on service delivery are real.

The temptation with any "no more custom integrations" claim is to treat it as a green light to scale agents fast. It isn't. Removing the integration bottleneck only exposes the next one: governance over what an agent is actually allowed to discover and act on. A customer-obsessed operator should treat MCP-style connectivity as a reason to tighten permissioning and audit trails before expanding agent scope, not after. The workflows an agent can now reach were often deliberately siloed for a reason — service design has to catch up to the plumbing, not the other way round.