Data Formulator 0.7 Bridges Fragmented Enterprise Data mit Agentic Workflows

Enterprise data workflows are notoriously fragmented, often trapped between disparate storage systems, BI tools, and isolated chat interfaces that lack persistent access to the underlying truth. Data Formulator 0.7 addresses this by providing an open-source, AI-powered system that connects these silos into a unified, agent-guided workspace.

Persistent Connectors Replace Manual File Uploads

Standard AI interactions often fail because they rely on manual, one-off file uploads that lack metadata, governance, and reproducibility. Data Formulator 0.7 introduces a Data Connectors feature that supports authenticated, persistent connections across databases, warehouses, BI systems, object stores, and local files. This architecture allows platform teams to manage reusable, governed connections once, rather than rebuilding integration logic for every individual analysis. Analysts can load, query, and visualize shared data directly within the workspace, ensuring that the AI agents are working with the same authoritative sources as the human users.

Agents That Reason Through Tools Rather Than Text

Unlike basic LLM prompts that attempt to solve complex problems through text alone, the agents in Data Formulator 0.7 operate within a shared workspace that includes connected data, loaded tables, and prior chart history. These agents reason and act through a suite of tools, allowing them to inspect data, execute code in isolated environments, and generate precise chart specifications. When a user's request is ambiguous, the agent is designed to ask clarifying questions before proceeding, which enables more complex, multi-step workflows such as transforming data, suggesting follow-up analyses, and generating batch visualizations. Crucially, every result is backed by verifiable, reproducible code.

Multimodal Workspaces for Iterative Exploration

Long-running analytical sessions are often lost in the linear scroll of a standard chat window. Data Formulator 0.7 utilizes a "Data Thread"—a structured chat that records every question, intermediate finding, and chart—allowing users to revisit earlier steps or branch into entirely new analytical directions without losing context. This is paired with an interactive canvas where users can directly edit visualizations or use natural language to instruct agents to adjust labels, annotations, and layouts. This combination of structured history and direct manipulation allows teams to move seamlessly from open-ended exploration to polished, shareable reports.

This release provides a foundation for teams to build more robust, agentic analytics systems that respect the complexities of enterprise data environments.

Source: Data Formulator 0.7: AI-powered data analytics for enterprise data
Domain: microsoft.com