Timeglass AI MCP

Timeglass AI MCP is the connector from Timeglass that links its automatically captured work record to the AI assistant a team already uses. It runs as a remote MCP server over Streamable HTTP with OAuth, so users can ask questions about projects, clients, meetings and timesheets in plain language.

What It Is

Timeglass captures work on screen and in meetings and drafts timesheets from it. The MCP connector exposes that record to assistants, giving them persistent context about minutes, tasks, meetings, projects and clients. Claude, ChatGPT and Gemini each have a documented setup path, and any assistant that supports remote MCP can connect to the same address.

How the connection works

Setup uses the assistant's own connector settings. The user adds the Timeglass address as a connector, signs in to Timeglass and approves the connection, then turns it on in a new conversation. The vendor states that no code, installation or configuration file is needed.

What you can ask and do

Example questions include why a project ran over its planned hours, what a person spent the day on, what was agreed on a client call, and how to build a playbook from how work is done. The assistant can also make changes when asked: it can add and edit time entries, create and update projects, clients and categories, and move activity to the right project. Entries created for someone else start as pending until that person reviews them.

Access and data rules

The connector follows the same access rules as the user's Timeglass account, so the assistant only sees what that role can already see in the app. Timeglass states that it does not charge a separate fee for the connector, and questions run on the user's own assistant plan.