Excel Copilot Adds GPT and Claude Selection
Geeky Gadgets reported on August 10 that Excel Copilot supports switching between GPT and Claude models for spreadsheet tasks, alongside natural-language analysis, formula generation, and live-data workflows. A Microsoft Community Hub post published in February states that Claude appears in Copilot's model selection only when an IT administrator has enabled Anthropic, while Copilot's Agent Mode can reason across worksheets and write and run Python.
Excel Copilot now offers model selection between GPT and Claude for spreadsheet work, according to reporting by Geeky Gadgets and a May walkthrough published by Finstory AI. The reported workflow places a model picker in the Copilot pane, where users can select Microsoft's default model, GPT-5.5, or Claude Opus 4.7.
Microsoft's Healthcare and Life Sciences Community Hub described a more qualified availability model in a February post: Claude appears as a Copilot option when an organization's IT administrator has enabled Anthropic. That distinction matters for enterprise deployments, where an administrator setting determines what users see in the application.
Spreadsheet automation and analysis
Geeky Gadgets describes Excel Copilot features including natural-language requests, automatic formula completion, live-data integration, reusable workflows, and change tracking. Its examples include requests to summarize monthly sales and highlight outliers, rather than requiring users to construct formulas manually.
The Microsoft Community Hub post focuses on Copilot in Excel's Agent Mode. According to that post, the feature can reason across multiple worksheets, write and run Python, structure data for dashboards and reports, and detect and correct some errors. The post demonstrates a synthetic paramedic-overtime dataset transformed into structured tables, visualizations, and an executive summary through a sequence of prompts.
Finstory AI reported testing Claude Opus 4.7 in Excel for a fictional 24-month SaaS forecast, then using Copilot to create a PowerPoint presentation from the work. That is a third-party demonstration rather than a Microsoft benchmark, so it does not establish general performance or reliability across financial models.
A broader Microsoft 365 integration picture
Anthropic separately advertises "Claude for Microsoft 365," with support for Excel, PowerPoint, Word, and Outlook. Its product page says Claude for Excel can answer questions about cells, update assumptions while preserving formulas, and build models across multiple tabs. Anthropic also describes review mechanisms such as highlighted Excel cells and tracked changes in Word.
The available materials therefore describe two related but distinct access paths: Claude as an administrator-enabled option within Microsoft 365 Copilot, and Anthropic's own Claude for Microsoft 365 offering. Organizations evaluating either path should verify licensing, tenant controls, data handling, model availability, and audit behavior rather than assuming that a Claude installation automatically enables model selection in Copilot.
For data and analytics teams, the practical question is less whether a spreadsheet assistant can draft a formula than whether its generated formulas, Python execution, source data, and workbook edits can be reviewed and governed. Comparable enterprise AI rollouts commonly require human validation for financial calculations, data-quality transformations, and externally sourced values, particularly when outputs feed operational reporting or decision workflows.
Key Points
- 1Excel Copilot model selection reportedly exposes GPT and Claude options, but Microsoft Community Hub notes Anthropic access depends on administrator enablement.
- 2Agent Mode can reportedly reason across worksheets and execute Python, extending spreadsheet assistance beyond formula drafting into data preparation workflows.
- 3Comparable enterprise deployments require review of generated formulas, executed code, data provenance, and workbook changes before operational use.
Scoring Rationale
Multi-model selection and agentic spreadsheet capabilities are notable for analytics teams that work extensively in Excel and Microsoft 365. The reported features could affect formula generation, workbook analysis, and Python-assisted workflows, although availability is administrator-controlled and the source materials do not provide independent reliability benchmarks.
Sources
Public references used for this report.
Practice with real Ad Tech data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Ad Tech problems
