MindsHub Cowork

MindsHub Cowork is an open-source application that provides a unified workspace for delegating research, analysis, and reporting tasks to AI agents. It allows users to connect various data sources and route tasks through different language models to generate shareable results. The software is available as a web application, a desktop app for macOS and Windows, or as a self-hosted installation on a local machine or VPC.

The platform functions as a superproject that integrates a frontend, an agent backend, and a data engine. Users can automate multi-step workflows, such as recurring monitoring or report generation, and create internal tools like dashboards and documents. These outputs, referred to as artifacts, can be published to a live URL for team distribution. This allows operators to turn raw agent output into structured assets that are accessible to a wider organization.

Key features

• Secure data vault for connecting BigQuery, Postgres, Gmail, Drive, HubSpot, Notion, and Linear
• Model router to switch between frontier and open models without individual provider keys
• Support for interchangeable open-source agent harnesses including Anton and Hermes
• Conversion of agent output into published documents, dashboards, apps, and code
• Cross-session memory and a reusable library of skills
• Scheduling system for recurring tasks and operations
• Flexible deployment options for cloud, VPC, on-prem, or air-gapped infrastructure

The architecture relies on a modular system of submodules for the core API, agent logic, and data vault. It is designed for knowledge workers, including strategists and creators, who need to build AI-driven analyses and tools without deep engineering requirements. The system manages credentials per connection to ensure agents do not have access to raw API keys, maintaining a layer of security between the data sources and the model execution. The build process supports various environments, including Electron for desktop use and a web SPA for browser access.

MindsHub Cowork is positioned as a flexible platform for managing AI agents and connected data across diverse deployment environments.