Until recently, working with AI for events felt rather like talking to someone very clever who had no access to anything. You pasted the programme into the chat, asked it to draft an email to the speakers, copied the response and added it to the platform yourself. AI helped, but moving information from one place to another was still your job.
That is changing, and at Eventscase, we already have the piece that makes it possible for our platform: our own MCP. With it, the AI your team uses can work directly within your events, adding, editing and deleting agenda items, attendees and speakers. And precisely because it can do all of that, we’ve decided not to release it on its own. First, we want to add one final layer: EVA Brain, which determines who can work on each project and holds the team’s shared memory.
In this article, we explain what an MCP is without getting technical, what it means for a team that organises events, and why we believe AI for events needs more than a connection to be genuinely useful.
What is an MCP, and what does it change about AI for events?
MCP stands for Model Context Protocol. Anthropic introduced it in November 2024 as an open standard for connecting AI assistants to the systems that hold data: workplace tools, document repositories and business applications.
The simplest way to understand it is to think of a universal plug. Previously, getting AI to work with a particular tool meant someone had to build a bespoke integration for that specific combination. With an MCP, the tool makes its capabilities available once, and any compatible assistant can use them.
From AI that answers to AI that acts
The practical difference is significant. AI without a connection answers questions about the information you give it. AI connected through an MCP can retrieve that information from the platform and, with permission, change it.
Take a familiar situation. Your client sends over the final programme in a document with twenty sessions, three room changes and two new speakers. Without MCP, AI can help you organise that document, but you still have to upload everything to the platform, session by session. With the Eventscase MCP, you ask your AI to build the agenda from that document, and it does so directly within your event.
A standard the industry already uses
This is not something we invented, nor are we taking an isolated approach. OpenAI adopted MCP in March 2025 and integrated the standard into its products, including the ChatGPT desktop application. In December 2025, Anthropic donated it to the Agentic AI Foundation, a Linux Foundation fund. This matters for a very practical reason: the Eventscase MCP works with the AI your team already uses, such as Claude or ChatGPT, without forcing you to switch tools.
What the Eventscase MCP will be able to do
In this first phase, the Eventscase MCP brings AI for events into three areas of any event: the agenda, attendees and speakers. Across all three, it can add new information, edit existing information and delete it.
For the agenda, that means building or updating the programme from the document your client sends, changing a session’s time or moving it to another room. These are tasks that currently take hours of copying and pasting, and where a mistake becomes visible as soon as the first attendee opens the event app.
For attendees, it means uploading a list, correcting details or removing someone who has cancelled. For speakers, it means adding biographies, updating job titles or removing someone who is no longer taking part. This information keeps changing until the last day and usually arrives by email, in a different format each time.
If you already use AI in your day-to-day work, as we discussed in AI for events: 8 practical applications, the shift here is that it stops being a separate tool and starts working with your event’s actual data.
The problem is not the connection — it is the team
When we built the MCP and saw it working, we asked ourselves an uncomfortable question: do we really want anyone to be able to connect AI to an event, give it permission to delete attendees and walk away? The answer brought us to two problems that have less to do with technology than with how events teams work.
Who can change what
An MCP connects, but it does not make decisions. If AI can read an event’s data, it can also delete it; if it can change a session, it can also change the wrong one. And it acts on behalf of the person giving it instructions.
For an events team, that is no small detail. In the weeks before a conference, extra staff join, people from the agency and the client work together, and each person needs access to a different part of the project. It is the same principle as event access passes: nobody gives every member of staff a pass that opens every door in the venue. AI needs to work in the same way, which is why every project needs clear rules about who can view information and who can add or change it.
Where does the team’s knowledge live?
The second problem is less visible and, for many teams, even more significant. Today, when someone uses AI to prepare an event, everything that AI learns lives in that person’s individual account: agreements with the client, why a supplier was ruled out, what changed in the last meeting. It is like a notebook everyone takes home at night.
While that person stays on the project, it works. But people move in and out of events teams all the time. Someone switches projects, extra help arrives three weeks before the event, or the next edition begins with a different person in charge, and what the AI knew is not available to anyone else. Everything has to be explained again. Or worse, decisions are made without knowing what has already been agreed.
Connecting AI for events to the platform without addressing these two issues speeds up the work, but it also speeds up the mistakes.
EVA Brain: the layer we’re finishing
EVA Brain is our response to those two problems, and the reason the Eventscase MCP has not been released yet. Think of it as the team’s shared brain: a place in the cloud that holds each project’s context and controls who can work with it.
EVA Brain gathers project information, remembers it and makes it available to the people working on that project. It stores each event’s agreements, changes and background in one place, without relying on any one person. It is also designed to allow an agent to join meetings, transcribe them and add what was discussed to the project, so a decision made on a call does not remain solely in the memories of the people who attended.
Each project retains its own context, separate from the others, so information about one conference does not get mixed up with another event for the same client. When someone joins or takes over, they can get up to speed with what is already known rather than starting from scratch.
Permissions and change history
EVA Brain’s second function is control. It determines who can view each project and who can add or change information. It also keeps a history of changes, so the team can review how its knowledge has developed.
This is where it fits together with the MCP. An AI that can delete attendees needs someone to decide what it is allowed to touch, and that is the layer we are finishing before launching the Eventscase MCP.
Working with your usual AI
From the user’s perspective, the process is straightforward. You connect EVA Brain to your usual AI, such as Claude or ChatGPT, through MCP. From then on, when you ask it to prepare a brief, proposal or report, it works with the project’s context and what the team has agreed along the way, rather than just what you tell it at that moment. It is a way to stop AI for events from starting from scratch in every conversation.
What changes for an events team
AI agents for events are often presented with a broad promise: AI will do things for you. We prefer to look at situations any team will recognise.
The first is bringing in extra staff during busy periods. Someone joins three weeks before the conference, and there is no time to explain everything that has already been decided. With EVA Brain, the AI that person uses works with the project’s context, but only the parts they have permission to access.
The second is the next edition. The team starts preparing next year’s event and wants to revisit what worked with suppliers, what problems arose, and what was decided and why. If that information was added to the project, it is still there, even if the person who contributed it has moved on.
The third is a change of project lead. When someone leaves a project halfway through, the project’s knowledge remains available to authorised team members, without relying on anyone’s memory.
In each case, the MCP enables AI to move from answering questions to taking action within Eventscase, while EVA Brain gives that action context and boundaries. It is the difference between AI agents for events that work on their own and those that work as part of the team.
Why AI in events needs time to become useful
We could have launched the Eventscase MCP as soon as it worked. It would have been quicker, and we could have said we got there before anyone else. But AI in events is only useful if the team can trust it, and that trust depends on two things a connection alone cannot provide: knowing what each person is allowed to do and holding on to what the team already knows.
AI for events will increasingly work inside the tools you already use, rather than alongside them. The question that sets teams apart will no longer be whether they use AI, but whether their AI knows what the team knows and respects what each person is authorised to decide.
That is where we want to be.