# Model Context Protocol (MCP) > Model Context Protocol (MCP) is an open standard that lets AI assistants connect to outside tools and data sources through a common interface. For a construction firm, it can let an AI agent use a project management system, CRM, or document library without a separate custom integration for each connection. - Stages: Preconstruction, Operations, Business Development - Concepts: Integration, Agents - Published: 2026-08-28 - Canonical: https://buildr.com/library/model-context-protocol ## Why it matters in construction A GC's data can live across the CRM, estimating package, project management platform, accounting system, document repository, and email. An agent asked "what is our exposure on the hospital job if the electrical sub slips two weeks" may need the schedule, subcontract, buyout log, and forecast. Before standards such as MCP, connecting an AI to each system often required a custom integration. With MCP, a system can publish its available tools in a common format for compatible AI clients. ## How it works MCP involves two sides. 1. **Servers** sit in front of a system and publish tools and resources. A project management server might expose "list open RFIs," "get submittal status," and "search documents." Each tool has a name, description, and expected inputs. 2. **Clients** are AI applications. When a user asks a question, the model sees the available tools, chooses which to call, and uses the results in its answer. Buyers should pay attention to three parts. - **Discovery.** The client learns what the server offers when it connects, rather than relying on a hand-coded tool list. - **Permissions.** The server controls what the connected user may see and do. Confirm that an AI client cannot exceed that user's access. - **Auditability.** Each call can be logged with its tool name and parameters. MCP is a connection layer. The agent still has to choose the right tools and interpret their results, and it can still be wrong. ## Example in practice For example, a commercial GC with about $180M in annual revenue runs BD out of a CRM, estimating in a dedicated tool, and active projects in a project management platform. A BD lead is prepping for a Thursday go/no-go meeting on a $22M higher-ed renovation. She asks the firm's AI assistant to build the go/no-go brief. Through MCP, the assistant pulls the opportunity record and last three touchpoints from the CRM, finds two comparable completed projects and their final margins from the project platform, checks the estimating tool for bid-day load that week, and pulls the owner's payment history from accounting. It assembles a one-page brief with each fact tagged to its source system. The assistant produces a brief with facts tagged to their source systems. Because it has read-only access, it cannot change any record without a person making the update. ## Go deeper - [Why MCP Is the Most Important Acronym Since BIM](/blog/mcp-construction-ai.md) - [How to Connect Your Preconstruction Workflow to Procore Without Manual Data Entry](/blog/connect-preconstruction-workflow-to-procore.md) - [Agentic: What It Means, and What It Means for Construction](/blog/what-is-agentic-ai-in-construction.md) ## Related terms - [AI Agent](/library/ai-agent.md): An AI agent uses a language model and connected tools to carry out multi-step tasks. For example, it can read a bid invitation, check it against go/no-go criteria, and create an opportunity in the CRM. - [Agentic Workflow](/library/agentic-workflow.md): An agentic workflow is a business process in which AI agents handle defined steps, such as intake, extraction, and record updates, while people review the work at checkpoints. A preconstruction team might use one for bid intake or CRM upkeep. - [The Lethal Trifecta](/library/lethal-trifecta.md): The lethal trifecta is a security pattern in which an AI system can access private data, read untrusted content, and send information out. In construction, a planted instruction in a bid or email can exploit a tool with all three capabilities and expose company data. - [Prompt Injection](/library/prompt-injection.md): Prompt injection is an attack in which instructions hidden in content an AI model reads, such as a subcontractor proposal or RFP attachment, alter the model's behavior. It is a major security risk for construction AI tools that process outside documents. - [AI-Native vs. AI Bolted-On Software](/library/ai-native-vs-bolted-on.md): AI-native software puts AI models inside its core workflows and data model. Bolted-on AI is usually a separate feature added to an existing product. For a construction firm, the difference affects whether the AI can work with pipeline, estimate, and staffing data or only discuss it. ## FAQ ### Is MCP an API? It is a standard way of describing and calling tools, built on top of APIs. A vendor exposes an MCP server, and any MCP-compatible AI client can discover what that server offers and use it. You still need the underlying system to have an API. ### Is it safe to give an AI agent access to our systems through MCP? It can be, with the right scoping. Start with read-only access, limit which tools the agent can call, log everything, and be careful when the agent can both read untrusted documents and take actions, since that combination is where prompt injection becomes a real risk. ### Which construction systems support MCP? Support varies by vendor. Ask whether a vendor exposes an MCP server, which tools it includes, and how it handles authentication and permissions.