What is MCP? (And why L&D and HR leaders need to pay attention)

What is MCP? 

You may have heard this term at a conference, Googled it, and still ended up confused. So let’s start with the thing that actually matters: MCP is a standard. Think of it like the plug socket on a wall. You don’t need to understand the wiring behind it; you just plug your devices in and go. 

But if you’re in L&D or HR right now, MCP is probably the most important acronym you’re not yet fluent in. That’s all about to change, because it’s already starting to reshape which learning platforms stay relevant, and which ones become invisible. 

‍

What does MCP actually stand for? 

MCP stands for Model Context Protocol. It’s an open standard created by Anthropic (the company behind Claude AI), meaning any technology company can adopt it. And they have. OpenAI, Google, DeepMind and Microsoft all followed within months of its launch in late 2024.

Here’s the simplest way to think about it. 

Cast your mind back to before the USB-C existed. Every device needed its own cable. Your phone had one connector, your camera had another,  your laptop had something different again. It was a mess, and plugging everything together required a lot of faff. 

MCP is the USB-C moment for artificial intelligence. It’s a universal connector that lets AI tools plug into the software systems a business already uses - HR platforms, document libraries, communication tools, and yes, learning management systems, without needing a custom-built connection for every single pairing. 

‍

"MCP is the protocol that turns your AI assistant from a very smart chatbot into something that can actually do things inside the systems your organisation runs on."

Before MCP existed, every company building AI tools had to invent their own way of connecting AI to other software. Custom integrations. One-off builds. Expensive, slow, and fragile. MCP changes that by making the connection standard, so any AI that speaks MCP can talk to any system that supports it. 

‍

Why does this matter for HR and L&D?

Think about what you actually do every day. You’re juggling learner data, content libraries, skills frameworks, onboarding workflows, performance reviews, and about twenty-five different browser tabs. The challenge for you has never been a lack of data; rather, it’s getting the information to the right person at the right moment – without everything becoming a full-time admin job. 

That’s exactly what AI, connected via MCP, aims to solve. 

Today, your AI tools mostly live in their own lane. You might use an LLM to draft an email. You use your LMS separately. You pull a skills report from a third tool. Everything works in isolation. 

With MCP, the walls between those tools start to come down. Your AI assistant can see what your learners are actually completing. It can check your HRIS to understand what roles are changing. It can surface the right content at the right moment, not as a separate task to go and find. 

‍

A concrete example of MCP in action

Let’s say a new manager joins your organisation. Without MCP-connected tools, getting them up to speed is a lengthy, manual process. Someone sends them a Confluence page, someone else adds them to a Slack channel, your L&D team manually assigns a few courses and hopes for the best. 

With AI connected via MCP, that friction is removed. The moment that new manager is added to your HRIS, everything else can follow. The right onboarding pathway appears in the LMS. Their calendar gets blocked for a ‘Manager Essentials’ session. Their team gets a heads-up. And if they have a question at 11pm on a Sunday, the AI can answer it from the context of their actual role instead of a generic knowledge base article. 

This is just the start of what MCP can make possible. 

But here’s the question L&D leaders will increasingly be asked 

If AI tools can reach into your LMS through MCP, why do you need a dedicated learning platform at all? Can’t people just use what’s already there, for example, the HRIS that tracks completions, the intranet with its document library, the SharePoint folder of PDFs? 

It’s a fair question, and we see versions of it in our own data. Some of the most common reasons organisations don’t push forward with a dedicated LMS comes down to familiar objections. The budget isn’t available, L&D isn’t seen as a current priority, and the existing platform is deemed good enough for now. These aren’t unusual responses, they reflect the real tension L&D leaders face: The case for investing in learning infrastructure can feel abstract when a business is focused on cost control and short-term performance. 

MCP changes that conversation because it makes the cost of not having the right platform much more visible. Why? Because MCP is a connector. It’s brilliant at what it does. But the value it delivers depends entirely on the quality and depth of what’s on the other side of that connection. 

An HRIS can tell an AI agent that someone completed a course on a given date, it cannot tell the AI what skills that course built, how they map to the employee’s current role, what the recommended next step in their development journey is, or how their progress compares to peers. A SharePoint folder of PDFs can be surfaced by an AI agent, but it cannot adapt based on who’s asking, what they already know, or what their manager has flagged as a priority. 

A purpose-built learning platform does all of those things. It holds the structure, the skills taxonomy, the pathways, the social learning signals, the completions, the assessments, and the recommendations that make AI-surfaced learning genuinely useful rather than just technically possible. 
But there’s a distinction worth knowing before you start conversations with your vendors. AI-enabled means the platform uses AI. AI-interoperable means your organisation’s AI can use the platform. Those aren’t the same thing, and in 2026, only one of them actually matters. 

‍

How MCP went from new release to industry standard

MCP was released publicly at the end of 2024. In AI years, that's ancient history — but by human standards, it just happened. And yet in that short window, every major AI company signed up. 

When the whole industry moves that fast in the same direction, it tends to mean one thing: This is where everything’s heading. 

Frankie Woodhead, Chief Product and Technology Officer at Thrive, has been clear about what this means in practice. Enabling customers to interact in the channel of their choice (web, mobile, APIs, and now MCP servers, is the next natural step. The learning platform doesn’t disappear behind the MCP, it becomes more accessible through it. MCP is how your learning content reaches people where they already are, rather than asking them to come to you. 
‍
Thrive’s MCP server is already in beta for 500+ customers, connecting the platform to AI agents from Claude, ChatGPT and more - covering skills including authoring and publishing content, managing social spaces, tracking user insights on assignments and completions, and endorsing content. 

For L&D and HR leaders, the timing matters. The organisations that understand what MCP enables, and start asking the right questions of their technology vendors, will be well ahead of those who catch up later. 

Worth knowing: MCP doesn’t replace your existing tools. It’s the layer that allows them to talk to each other through AI. Think of it less as a new platform, and more as the connective tissue your tech stack has been missing. 
‍

What does this mean in practice for your tech choices? 

When you're evaluating any AI-powered learning or HR tool from here on, MCP support is worth adding to your checklist. Ask:

• Does this platform support MCP, or does it have a clear roadmap to get there?
• Can it connect to the other tools in our stack e.g. our HRIS, comms platforms, and AI assistants?
• Can the AI act on information across systems, or does it only know what's inside its own walls? 

The platforms that support open standards like MCP are the ones being built for where work is heading, not where it's been. 

‍

‍

From understanding MCP to acting on it

MCP is the open standard that makes AI useful inside the systems your organisation actually runs on. It’s why your AI assistant can go from answering questions to doing things: Surfacing the right content, triggering the right workflows, and joining the dots between your tools. 

You don't need to become a developer to get value from this. But you do need to know what questions to ask. 

For L&D teams who get ahead of this, who understand what MCP enables and start asking the right questions of their technology vendors, the next budget conversation looks very different. Learning stops being the thing employees have to go and find, it becomes the thing that finds them, and that’s a pretty good place to be. 

This is the first in Thrive's MCP series for L&D and HR leaders. Next up: why your learning platform is probably the most disconnected tool in your AI stack — and what that's costing you.

Thrive is the modern learning platform built for the way people actually work. Our MCP Server is now in beta, connecting Thrive to the AI tools your teams already use. Want to see how it works? We’d love to show you. 

‍