OpenDashboards launches open source tool for private AI dashboards

Conversational dashboard building

OpenDashboards is a new open-source application designed to create data visualizations through natural language. Instead of manually configuring charts and filters, users can simply ask a question to have a dashboard built automatically. This approach aims to provide a single application that can generate an infinite number of different dashboards based on specific needs.

The system is designed to pull information from various sources simultaneously. When a user asks a question, the tool identifies the necessary data points and creates a dashboard that aggregates that information into one view. This allows for a more fluid way of exploring data without having to switch between different software tools or manual reports.

A key feature of this platform is its efficiency in updating information. When a user selects a specific filter, every widget on the dashboard updates to reflect that choice. If new data is added to the source, the dashboard remains the same without needing to be regenerated. Furthermore, if a user chooses to change only one specific chart, the system only regenerates that single component rather than the entire layout.

Data sources and connectivity

The platform supports a wide range of popular databases and analytics tools out of the box. Users can connect to systems such as Postgres, MySQL, ClickHouse, BigQuery, PostHog, Google Analytics, and Search Console. These integrations allow the tool to read across different data silos to provide a unified answer to user queries.

For organizations using specialized software that is not natively built into the system, OpenDashboards utilizes the Model Context Protocol (MCP). This protocol allows the tool to connect to external services like Stripe or HubSpot. By using MCP servers, the platform can extend its reach to almost any tool that provides a compatible connection point.

The architecture ensures that the underlying data remains where it is currently stored. OpenDashboards reads across these various sources to provide answers without moving the primary data into a new location. This design is intended to help users maintain their existing data workflows while still gaining the benefits of an AI-driven visualization layer.

Hosting and privacy options

Privacy is a central component of the OpenDashboards offering, as it provides multiple ways to deploy the software. Users have the choice to use the hosted version at opendashboards.com or to self-host the application on their own private servers. Self-hosting allows the entire app to run inside a company's own network, which is a critical requirement for many enterprises handling sensitive information.

When a user chooses to self-host, their credentials stay on their own server, and all queries are run from that local environment. This setup ensures that the sensitive keys and database access points are not exposed to external third parties. The app remains open source, which allows for greater transparency regarding how the software handles data and interacts with the user's infrastructure.

For those who prefer not to manage their own servers, the hosted version provides a managed experience. However, for complete isolation, the OpenUI Cloud can be privately hosted inside a private network. This gives organizations the flexibility to choose the level of security and infrastructure management that best fits their specific operational needs.

Model flexibility and APIs

While the tool offers broad flexibility in hosting and data, there are specific requirements for the AI models used. The platform builds dashboards using the OpenUI Dashboards API, which runs on OpenUI Cloud. This API is responsible for building, storing, and updating the dashboards and chats generated by the users.

Users can choose to use various models through the OpenUI Dashboards API. The platform also supports Bring Your Own Key (BYOK), which allows users to add their own OpenAI or Anthropic keys at the console. This means that while the app is open source, the underlying intelligence can be powered by the specific providers that a user already trusts or uses for other tasks.

Developers who are building their own products can also interact with this technology. The OpenUI Dashboards API can be embedded into other products to provide dashboard-building capabilities. OpenUI Cloud also offers a generous free tier for those who want to experiment with the API before committing to a larger deployment or a private hosting setup.

Claude integration limits

It is important to note that certain features are restricted to specific environments. Specifically, the Claude Dashboards only work inside Claude, using Claude's models on Anthropic's cloud. This distinction is important for users who may be trying to use the tool with different model providers or entirely different cloud environments.

The distinction between the open-source app and the specific Claude integration highlights the different ways the tool can be used. The open-source version is designed for broad compatibility with various models and self-hosted environments. In contrast, the Claude-specific dashboards are tailored for the Anthropic ecosystem, providing a more integrated experience for those already using Claude's services.

Users looking to get started can join a Discord community to ask further questions about these different configurations. Whether a user wants a fully private, self-hosted setup with their own keys or a managed experience using the hosted platform, there are different paths available. The choice depends on the user's specific needs regarding data privacy, model preference, and technical overhead.