MCP Gives AI the Tools. Expertise Teaches the Craft.
I have a lot of woodworking tools—saws, drills, routers, sanders, and probably more than I really need. But owning those tools doesn’t make me a craftsman. Put the same tools in the hands of someone who has spent 20 years building furniture, and the outcome will be very different.
The difference isn’t access. It’s knowing which tool to use, when to use it, how to use it, and what to do when something doesn’t go according to plan.
AI is reaching a similar point.
Today, we’re announcing MCP support in Bricklayer, enabling customers to extend Subject Matter Expert Agents with MCP-enabled product capabilities alongside Bricklayer’s existing integrations.
The Model Context Protocol has quickly emerged as an important standard for connecting AI systems to enterprise software. Just as importantly, its authorization model has matured, making the protocol increasingly suitable for enterprise environments. That combination makes now the right time for Bricklayer to support MCP.
But connectivity is only part of the problem.
Access to a Tool Is Not the Same as Mastery of a Tool
APIs gave software access to other software. MCP takes an important next step by giving AI a standardized way to discover and interact with those capabilities.
What MCP doesn’t provide—and isn’t intended to provide—is the experience required to use those capabilities well.
Consider a security analyst working with a sophisticated security platform. Knowing that a search capability exists is very different from knowing how to structure the right search, progressively narrow an investigation, identify trustworthy data sources, recognize a suspicious result, or determine what to try when the first approach doesn’t work.
The same is true across enterprise software. Products provide capabilities. Expertise turns those capabilities into useful outcomes.
That distinction has influenced how we’ve built agents at Bricklayer from the beginning.
Building Agents Around Roles and Expertise
There are two fundamentally different ways to give AI access to an expanding collection of enterprise tools.
One approach is to give every agent access to every product it might need and then teach each agent how to use those products. A SOC Analyst Agent needs to understand every security platform it might touch. An Incident Responder needs much of the same product knowledge. So does a Threat Hunter.
As the environment grows, that knowledge gets duplicated across more agents. Product changes must be incorporated in multiple places, and each agent becomes responsible for an expanding collection of implementation details.
Bricklayer takes a different approach.
We separate Role-Based Agents (RBAs) from Subject Matter Expert Agents (SME Agents).
Role-Based Agents understand the work to be done. A SOC Analyst Agent knows how to investigate an alert. An Incident Responder knows how to coordinate an incident. A Threat Intelligence Analyst knows how to research and assess a threat. They determine the objective, understand the context, and decide what should happen next.
SME Agents come in three types: API, Datastore, and MCP. Each provides specialized knowledge and access for a particular system or data source. When an RBA needs a specific capability, it relies on the appropriate SME Agent for the relevant terminology, workflows, sequencing, limitations, and proven practices.
The RBA directs the work. The SME Agent contributes specialized expertise. API, Datastore, and MCP integrated agents connect the RBA to the capabilities and information it needs
Why the Separation Matters
This might sound like a subtle architectural distinction, but it becomes increasingly important as an agentic workforce grows.
When product expertise is embedded independently into every agent, knowledge gets duplicated throughout the system. Agents may learn different ways to use the same product, changes must be propagated across multiple places, and each new integration increases the amount of knowledge every agent is expected to carry.
Separating Role-Based Agents from SME Agents changes that equation.
Expertise becomes reusable. Build product expertise once, and every Role-Based Agent that needs that capability can benefit from it. Improving an SME Agent improves execution everywhere that expertise is used.
Maintenance becomes simpler. When a product changes, its SME Agent can evolve with it. You don’t have to find and update every agent that was independently taught how to interact with that product.
Execution becomes more consistent. Different Role-Based Agents aren’t inventing their own ways to use the same technology. They rely on a common body of product knowledge, workflows, and proven practices.
Expansion becomes faster. Adding another technology doesn’t require retraining an entire agent workforce. You add the appropriate SME Agent and make that expertise available to the roles that need it.
Organizational knowledge compounds. Every improved workflow, new technique, lesson learned, or product capability can be incorporated into an SME Agent and reused across the workforce. Expertise stops being repeatedly recreated and becomes an asset the organization continually improves.
This is a major part of what we believe an agentic workforce should look like: roles remain focused on the work, while specialized expertise becomes reusable infrastructure.
From Access to Expertise
APIs made product capabilities programmable. MCP gives AI systems a common way to discover and access those capabilities. Bricklayer’s SME architecture adds the next layer: the accumulated product knowledge required to turn access into reliable outcomes.
We believe the next generation of enterprise AI will be distinguished not by how many tools it can reach, but by how effectively it can capture, improve, and reuse expertise.
MCP in Bricklayer
With this release, Bricklayer customers can extend SME Agents with MCP-enabled capabilities alongside the integrations already available in the platform.
We’re also working with technology partners — including Microsoft, SentinelOne, ServiceNow, Sumo Logic, Wiz, and others — on MCP-driven initiatives. The goal is to pair the capabilities partners make available through MCP with the product-specific expertise needed to apply those capabilities effectively within an agentic workforce.
For us, MCP support isn’t another integration method to add to a checklist. It expands the capabilities Bricklayer’s agent workforce can use while preserving an architectural principle we believe matters: access to a product and expertise in using that product are two different things.
MCP gives AI a better way to reach the tools.
Bricklayer gives AI the expertise to use them well.
