Connect WebWork to AI Assistants with MCP
WebWork now offers a Model Context Protocol (MCP) connector, letting AI assistants work with your WebWork data through natural language. It works with any MCP-compatible AI assistant, such as Claude or ChatGPT, using a single connection. Every request runs as the signed-in user, respecting the same roles and permissions already in place. This is for workspace Owners and Executive Managers who want to manage projects, time, and approvals by simply asking, rather than navigating menus.
What You Can Ask
Ask in plain language and your assistant handles the rest. A few examples:
– Projects and tasks – list projects, review open tasks, or create and assign tasks to team members.
– Time tracking – check what someone is tracking right now, start a timer, or add manual time entries.
– Approvals – view pending timesheets, approve them in bulk, or reject a leave request with a reason.
– Reports – pull tracked-hours reports, activity reports for a date range, or attendance summaries.
– People – list workspace members or invite new users with a specified role.
How It Works
Ask your assistant in plain language and it calls the right action on your behalf. The connector acts as you, so it can only do what your role allows – anything outside your permissions is simply refused. It stores no separate copy of your data, and you can disconnect at any time to revoke access.
Permissions and Safety
- – Scoped access – grant read, write, and/or admin access at sign-in, and approve individual actions in your AI assistant.
- – Confirmation on destructive actions – deletes and archives require explicit confirmation before they run.
- – Transparency on emails – actions that send real emails, such as invitations and leave approvals, are flagged before they are sent.
- – Revoke anytime – disconnect at any point to immediately revoke access.
Ready to connect? See the setup documentation to get started.
