There's a distinction worth making between two different things AI tools can do with your business.
The first is answering questions about you: what you do, where you're located, what people say about you. AI models do this now, drawing on their training data and, for tools like Perplexity, live web searches. The quality of those answers depends on how visible and well-described you are across the web.
The second is interacting with you in real time: checking your actual availability, retrieving your current prices, completing a booking. This is different in kind — and it's where a standard called the Model Context Protocol is pointing.
What MCP actually is
The Model Context Protocol is an open specification, developed by Anthropic and now being adopted across the industry, that defines how AI agents connect to external data sources and services. It's not a product you buy or a platform you join. It's more like an agreed convention — the same idea as robots.txt or structured data markup, but for live, two-way connections.
A business that runs an MCP server is making certain information or functions available for AI agents to call. A spa might expose its appointment availability. A retailer might expose current stock levels. A professional services firm might expose its service descriptions and booking calendar. The AI agent — whether that's Claude, ChatGPT, or eventually a voice assistant — can query those things directly, without relying on whatever the model happened to learn during training.
The practical difference
Consider the gap between these two interactions:
"Is Serenity Spa good for a massage?" — an AI model can answer this from its training data and web knowledge. The answer might be months out of date. It might be wrong about the services on offer. But it can give a response.
"Does Serenity Spa have availability for a 60-minute massage next Tuesday afternoon?" — this requires live data. No amount of training or web crawling gives an accurate answer here. Without a direct connection to the spa's booking system, the AI has to say it doesn't know, or guess.
MCP is the mechanism that closes that gap. The spa's booking system is connected. The AI asks. The system answers.
Where this is now
It's worth being honest about the current state. MCP today is largely a developer tool. Connecting it to an AI assistant requires manual configuration — editing files, pointing to server addresses. The businesses experimenting with it are almost all in the technology sector, and the experience for end users is not yet seamless.
There are also early signs of what comes next. Work is underway on well-known endpoint conventions — the idea that a business could publish an MCP connection point at a standard URL, the same way they publish robots.txt, and AI agents could discover and use it automatically without any manual setup. Registries of available MCP servers are emerging. Remote connections over standard web protocols are becoming more practical.
The current phase is genuinely early. But the direction is clear: making it progressively easier for AI agents to work with live business data, and progressively less reliant on human configuration to make that happen.
What this means in practice — eventually
For most small businesses, MCP isn't something to act on today. The tooling isn't mature enough, the integrations aren't widespread, and the user behaviours that will make it valuable are still forming.
What it does represent is a new dimension of what it means to be visible to AI. Right now, visibility means being mentioned — appearing in the answers AI models give when someone asks about your category. That matters, and measuring it is useful.
The next layer is being queryable: having live information that AI agents can retrieve accurately when someone moves from research to action. Booking an appointment, checking a price, confirming availability. That transition — from passive mention to active interaction — is what MCP is designed to enable.
The two things are related. Being consistently mentioned in AI responses is, in part, what makes a business's name known to users who might then ask to interact with it directly. Visibility is the prerequisite for queryability.
For now, the most useful thing most businesses can do is understand where they stand on the first layer — how consistently AI models mention them, across which tools, for which queries. The infrastructure question about real-time connections will follow as the standards mature.
Run a free V-Score audit to find out how visible your brand is to the four major AI models today.