# Data Formulator 0.7 adds persistent enterprise data connections > Microsoft Research’s analytics release combines reusable data connectors, context-aware agents and a workspace for branching visual analysis. Canonical URL: https://www.devobs.io/news/news-data-formulator-0-7-release/ By: Nina Patel Published: 2026-09-06T11:58:54.630Z Updated: 2026-09-06T11:58:54.630Z Event date: 2026-05-28 Section: AI Microsoft Research released Data Formulator 0.7 on May 28, extending its open-source analytics system with reusable data connections and a workspace for iterative analysis. The [release overview](https://www.microsoft.com/en-us/research/blog/data-formulator-0-7-ai-powered-data-analytics-for-enterprise-data/) combines connection management, agent-assisted exploration and visualization editing. ## Analysis retains its working context The Data Connectors feature supports persistent connections across several data-source types. Context-aware agents can inspect the workspace, prepare data and generate charts. A Data Thread records analytical steps and supports branching into alternative investigations, while an interactive canvas lets users refine visual presentation. Microsoft positions the release as a foundation teams can adapt to their own enterprise requirements, rather than a single chat response that disappears from the workflow. ## Keep the calculation reviewable Our suggested adoption test starts with a question whose correct answer is already known. Follow the generated analysis from source rows through transformations to the displayed chart. A convincing graphic should not substitute for checking a join, aggregation or filter. Include ambiguous business terms in the test. If a request asks for “active customers,” the application should obtain or expose the definition rather than silently choose one. Keep that definition visible beside the result so a later viewer understands what was counted. ## Test the shared-workspace boundary Reusable connections reduce repeated setup, but teams should still verify which data each user and agent can access. Use a test account with deliberately limited permissions and inspect whether previews, saved results and shared analyses respect that boundary. For a useful pilot, ask a second analyst to reproduce an earlier conclusion from the recorded steps. The practical value of an AI analytics workspace is not only producing a chart quickly; it is helping another person understand and check how the chart was made. ## Source references -