Technology loves two things: 1) standards and 2) acronyms. MCP or "Model Context Protocol" is an emerging concept that in addition to firmly placing a tick against both of these loves is also starting to generate some interest.
But what is it?
We have been working with a building company to help better understand resource planning and work projects in relation to delays, availability of skills and equipment, changes to requirements and mother nature, the weather. Think of Sunday afternoon and looking at the week ahead, we have a plan, I have told my team where to be, when and what to work on. The question you want the answer to is "Does the weather forecast for tomorrow and the rest of the week affect my projects and resource planning?"
From an AI perspective, after you state the question to an AI assistant like Claude or ChatGPT, there are two outcomes.
Here is the information you wanted.
Here is the information you wanted. Do you want me to update the plan and notify the team on these changes?
Unsurprisingly, 2 is harder to do than 1.
Moving From Information to Action
Getting the information is one thing but turning that insight into action is another challenge entirely.
The problem? Getting AI assistants like ChatGPT or Claude to communicate with all these different operational systems can be difficult. Each system has its own specifics in relation to how it operates and processing of instructions.
MCP. Making Systems Friendlier to AI
This is where emerging standards like MCP come in. Similar to how APIs have helped systems talk to each other for years, MCP aims to make it easier for AI assistants to interact with these systems. The protocols are being developed to help action oriented systems become more accessible to AI tools like Claude and ChatGPT.
In essence, MCPs are designed to provide context around what the AI is asking for and then help execute those requests in the appropriate systems.
Where are we now?
It's worth noting that while the concept is promising, we're still in early days. The emphasis on developing MCP’s falls on the systems themselves as opposed to the AI assistants and is subject to love number 1, landing on a recognised and accepted standard approach and then the willingness to do so.
Why This Matters
Despite being an evolving standard, this represents an important step forward. As AI moves from simply answering questions to taking actions on our behalf, having standardised ways for AI to communicate with business systems becomes crucial.
For our building client, the vision is clear. Moving beyond weather alerts to systems that can automatically reschedule work, reallocate resources, and update project timelines - transforming how they respond to change.