Microsoft open-sources Data Formulator 0.7 for enterprise AI analytics

Microsoft has quietly opened the doors to Data Formulator 0.7, an open-source AI-powered system built to untangle the mess of enterprise data scattered across storage systems, tools, and teams. The release goes after three stubborn headaches in enterprise analytics: connecting governed data sources, providing context-aware assistance, and giving teams a shared workspace where they can refine their work.

Data Formulator 0.7 ships with a set of features that collectively slash the integration burden on platform teams while letting analysts and domain experts move faster. The code is already on GitHub, ready for organizations to customize and deploy as they see fit.

Connecting enterprise data with Data Connectors

The headline feature is Data Connectors, a new capability that allows governed, reusable links to a wide variety of data sources, databases, data warehouses, BI systems, object stores, and local files among them. Authentication, persistent connections, previews, and metadata management all live inside a single workspace, killing the need to manually upload files or rebuild connections from scratch for each analysis.

The payoff is less grunt work for platform teams and cleaner governance from the start, since users pull from centrally managed connections. Reproducibility becomes baked in rather than an afterthought.

Context-aware agents for data analysis

At the heart of Data Formulator are its context-aware AI agents, which have full access to the analysis workspace. Unlike the typical single-shot prompt, these agents can peek at connected data sources, loaded tables, earlier charts, and the user's stated goal. They reason and act through tools, not just text, which unlocks more complex analytical workflows.

In action, an agent can inspect data, write and run code in an isolated sandbox, produce chart specifications, and explain its results while showing intermediate steps. If a request is fuzzy, the agent stops and asks for clarification before charging ahead. That means agents can align analyses with the user's intent, prepare and transform data, suggest follow-up questions, generate tables and charts in bulk, and produce verifiable, reproducible code for every result.

A workspace for iterative data analysis

Data Formulator couples these agents with a multimodal interface designed for open-ended analysis. Users talk to agents through the Data Thread, a structured chat that logs every question, intermediate finding, and chart created along the way. Long sessions stay navigable, letting users revisit earlier steps, branch into alternative analyses, and compare them side by side without losing context.

The interactive canvas sits alongside the Data Thread, letting analysts tweak visualizations directly. When the task shifts from exploration to communication, they can refine charts on the canvas or describe changes in plain English and let the agent handle labels, annotations, layout, color, and emphasis. Teams can also generate reports and share their findings externally.

Implications for enterprise analytics

The arrival of Data Formulator 0.7 points to a broader shift in how enterprise teams can approach AI-assisted analytics. By bundling governed data connectivity, context-aware agents, and an iterative workspace, the system tries to close the gap between fragmented data and reproducible analysis. And because it is open source, organizations can tailor the tooling to their own systems and requirements.

Microsoft has posted a demo and the full GitHub repository for teams to kick the tires. As enterprise data workflows grow messier and more complex, tools like Data Formulator might just cut the friction of moving from raw data to real insights.