Microsoft Research releases Flint-interactive, a language for building interactive data visualizations

Microsoft Research released Flint-interactive, a language designed to simplify the creation of interactive data visualizations. The language allows users and AI agents to define behaviors using high-level semantic events and update operators, rather than writing low-level visualization code.

Flint-interactive relies on Flint’s compiler to translate user interactions into semantic events and apply high-level operators to update visualizations. This shifts the burden of resolving complex dependencies between interaction and visual logic from the developer to the compiler, improving reliability and reducing development effort.

The system supports interactive data stories, dashboards, and connected experiences where actions in one view can update charts, documents, widgets, or application state in another. For example, brushing one chart can highlight related groups of countries using a preset interaction like linked-brush.

Flint-interactive can be used directly in chat environments via Flint-MCP and MCP Apps, enabling agents to generate an interaction_spec alongside a Flint chart spec and render interactive charts in conversation.

Developers can also build interactive applications with Flint Interactive Canvas, using event handlers to process semantic events and update operators to control how charts respond to user input or external widgets.

For bespoke interactions beyond Flint’s presets, developers can define triggers and custom handle functions using a programming API. This allows for tailored interactions, such as dragging a point along a time path to visualize changes in multiple cities’ temperature and precipitation.

The approach addresses challenges in generating high-quality interactive visualizations, including high generation costs from repeated rounds of reasoning, coding, and debugging, as well as reliability issues like poor layouts or broken interactions.