We are developing an AI assistant for meeting management

At X Mind Solutions, we have started developing an AI assistant to support our own meeting processes. For now, this is an internal hobby project, and our goal goes beyond creating a tool that simply takes meeting notes. We are designing a virtual colleague that prepares information before meetings, organizes files and collects feedback afterwards. This approach treats meeting management as an integrated workflow, from preparation through to evaluation.

Developments such as Google’s Project Astra and Agent Mode were part of our motivation for starting this project. We are also focusing on MCP, or Model Context Protocol, as a way for the assistant to connect with different tools. Within the project, we are exploring these technologies with the aim of taking an assistant beyond simply answering questions. For us, the central challenge is to develop a system that can not only present information but also carry out related tasks within the relevant systems.

Our first goal is to be able to delegate the necessary preparations to the assistant before a meeting begins. We want it to research and compile brief background information and documents on the meeting topic, then organize the files. Creating a folder in Drive and gathering presentation files in it are also part of the workflow we are designing. In this way, we want the assistant to do more than suggest content: it should make the materials needed for meeting preparation available in an accessible, organized format.

After meetings, we aim to collect feedback from participants through a short survey. Another focus of the project is to analyze this feedback across meetings and use it to offer suggestions for more efficient meetings. The aim here is not simply to record what was discussed, but to make the meeting experience itself open to evaluation. We are not claiming any efficiency gains at this stage; the capability we want to develop is the ability to turn team assessments into insights that can inform future meetings.

We are aiming to build a system that can connect to tools such as Slack, Notion, GitHub and Drive through MCP. Automatically sending messages to Slack and creating pages in Notion are among the tasks we are considering. With these connections, we want the assistant to move from passively presenting information to actively managing tasks. At this stage, we see the work not as a finished product, but as an internal experiment developed around our own needs.