Agentic Distribution Moves To The Protocol Layer

Opening Thesis

The agentic economy is becoming an interoperability race.

That is the shift for today. The question is no longer just whether a company has an AI agent. The question is whether that agent can work with the systems, tools, data, and other agents around it.

This matters because agentic value rarely lives inside a single model. A sales agent needs CRM context, pricing rules, contract data, product documentation, and calendar actions. A commerce agent needs catalog data, inventory, payment rails, refund policies, and checkout flows. A support agent needs identity, ticket history, entitlement data, knowledge base content, and escalation paths.

If each agent needs a custom integration for every tool and every other agent, agentic adoption stays fragmented. If the market standardizes how agents access tools, exchange tasks, and coordinate across vendors, the distribution layer changes.

Yesterday's brief argued that agents expose thedata readiness gap. Today’s issue looks at the next step: once the data is usable, companies need a governed way for agents to reach it, act on it, and collaborate around it.

Strategic takeaway: the next agentic moat is not only the agent. It is the set of interfaces that make the agent useful across the business.

Signal 1: MuleSoft Pushes A2A Into Enterprise Orchestration

MuleSoft announced agentic innovations built for speed, governance, and scale, including anA2A Bridgegoing generally available in July 2026. The core idea is practical: companies already have agents built on different platforms, and many of those agents are not natively A2A-compliant. A bridge can translate protocols, map identities, manage task lifecycle state, and bring those agents under a centralized orchestration layer.

That is a business signal, not just an integration feature. It suggests the market is moving from “deploy an agent” to “coordinate a portfolio of agents.” In a real enterprise, no single agent will own every workflow. There will be sales agents, finance agents, commerce agents, IT agents, service agents, procurement agents, and analytics agents. They will need to hand off work without forcing every team to rebuild the same integration layer.

For founders and CMOs, the implication is that agentic maturity will become visible in how products participate in orchestration. If your product contains useful data or actions, buyers will ask whether agents can reach it through standard interfaces. If your company claims to use agents internally, investors and operators will ask whether those agents are isolated demos or part of a managed operating system.

Strategic takeaway: enterprises will not scale agents one assistant at a time; they will scale them through orchestration.

Signal 2: MCP And A2A Are Becoming Different GTM Surfaces

Celigo’s recent explanation ofMCP vs A2Ais useful because it makes a distinction growth teams need to understand. MCP governs how an agent accesses tools, data, and business systems. A2A governs how one agent coordinates with another agent. They solve different business problems.

That distinction matters for go-to-market strategy. MCP is about making your company callable. Can an agent retrieve your pricing? Search your catalog? Check availability? Create a ticket? Start a quote? Pull analytics? Update a workflow? If yes, your product or brand becomes easier for agents to use.

A2A is about making your company coordinatable. Can your specialist agent receive a task from another agent, return status, hand off context, and participate in a multi-step workflow? If yes, your company can plug into broader agentic workflows where another platform owns the user relationship.

The Linux Foundation’sA2A milestoneshowed why this matters. The protocol crossed 150 supporting organizations and reported production use across industries including supply chain, financial services, insurance, and IT operations. That means agent coordination is moving from research novelty to enterprise architecture.

For founders and CMOs, the practical question is simple: what should your business expose through MCP, and where should your business participate through A2A? Answering that is now part of distribution strategy.

Strategic takeaway: MCP makes you usable by agents; A2A makes you useful inside agent networks.

Signal 3: Commerce Shows Why Protocols Become Revenue Infrastructure

TechRadar’s latestagentic commercepiece argues that AI agents are transforming ecommerce by interpreting intent, navigating discovery, and executing purchases in real time. It also points to the infrastructure shift underneath: agentic commerce needs secure intent verification, authenticated agents, and programmable payment flows.

Salesforce is already showing the concrete commerce version of this shift. ItsAgentic MCP Shopper Toolspilot is designed to connect AI agents such as Claude, Agentforce, and ChatGPT to real-time B2C Commerce storefront data. The documented service supports shopper-focused tools and references MCP, A2A, and secure authentication as part of the standards-based approach.

This is why protocols matter commercially. A product page can persuade a human shopper. A protocol-connected commerce layer can let an agent search, compare, assemble a cart, and move toward checkout using live storefront data. That changes who controls demand.

For commerce leaders, this is not a future-site redesign question. It is a distribution design question. Which product facts should agents retrieve? Which cart actions should they prepare? Which payments require explicit human approval? Which marketplace or assistant partners should be able to call the storefront? Which agent actions should be logged for disputes?

This connects directly to the recent brief onstorefront infrastructure. The storefront still matters, but the callable layer around the storefront may matter just as much.

Strategic takeaway: agentic commerce turns protocols into revenue channels.

What To Do This Week

Create a protocol map for one important business workflow.

Start with the workflow, not the technology. Pick one high-value flow: product discovery, quote creation, customer onboarding, support escalation, procurement, renewal management, campaign operations, or checkout.

Then list what an agent needs to access. Mark the systems, data, documents, APIs, product feeds, analytics, pricing rules, and action endpoints required to complete the workflow.

Next, separate tool access from agent coordination. Tool access is where MCP may matter: agent to system, agent to data, agent to action. Coordination is where A2A may matter: agent to agent, handoff to specialist, status return, task lifecycle, escalation.

Then define governance boundaries. Which calls are read-only? Which mutate data? Which trigger payments, messages, approvals, refunds, or customer-facing actions? Which require human confirmation?

Finally, translate the map into GTM assets. Your website, docs, sales materials, and partner pages should explain what your product exposes to agents, what standards you support, what actions are possible, and what controls protect the customer.

The practical move is to pick one workflow and decide what should become agent-accessible, what should become agent-coordinated, and what should stay human-controlled.

Closing Line

In the web era, distribution meant being reachable by browsers and search engines. In the agentic era, distribution will mean being reachable by agents, tools, and other agents that can safely act.