· Saul Mateos · Claude and AI Tools · 4 min read
Claude for Excel: The CFO's Guide to Claude Integrations
Claude for Excel, connectors, plugins and skills, explained for finance pros: what works today, what's still clunky and what to set up first.
Here’s what changed with Claude for Excel and the integrations around it: Claude went from “smart assistant you copy-paste into” to “assistant that lives inside your actual workflow.” It reads your spreadsheets, pulls market data, searches your email, and builds the deck. In one session. Without you re-explaining context every time you switch tabs.
I’ve been testing these. Some of them genuinely changed how I work. Some of them are still clunky. The CFO’s Guide to Claude Integrations is the honest breakdown.
Download the Claude Integrations Guide for Finance Pros (PDF)
What’s inside the guide
- What Happened While You Were Closing the Books
- The Three Layers: Connectors, Plugins, and Skills
- How to Set Them Up
- The Connectors That Matter for Finance
- The Skills That Change Everything (Financial Analysis, Investment Banking, Equity Research, Private Equity, Wealth Management)
- Claude in Excel and Data-to-Deck
- Bonus: Accounting Software via MCP
- The Honest Take
- Your Monday Morning Action
Claude for Excel and data-to-deck
Claude now has a sidebar inside Excel. It reads your workbook, analyzes the data, and can modify cells. But the real feature is the cross-app workflow: Claude carries context from Excel to PowerPoint without losing the thread.
The “data-to-deck” workflow: analyze earnings data in Excel, then tell Claude to build the summary slides in PowerPoint. It doesn’t ask you to re-explain the analysis. One session, two apps, zero re-pasting.
Data partners integrated into this workflow include S&P Global Capital IQ, LSEG, Daloopa, PitchBook, Moody’s, and FactSet.
Current status: Research preview on Mac and Windows. It works, but expect occasional rough edges. I’ve had it lose context once in a complex multi-sheet model. Save your work.
The three layers: connectors, plugins, and skills
Everyone’s calling these “integrations,” but there are actually three distinct things happening here. Understanding the difference matters because it changes how you set them up and what you get out of them.
Connectors are the pipes. They wire Claude to external tools and data sources: Google Drive, Gmail, Stripe, FactSet. A connector doesn’t make Claude smarter. It gives Claude access to your stuff so you stop copy-pasting between tabs.
Plugins are the packages. Each plugin bundles connectors, skills, slash commands, and sub-agents for a specific job function. The Financial Analysis plugin, the Investment Banking plugin, the Equity Research plugin. Think of them as role-based AI teammates that come pre-loaded with the right tools and expertise for a particular kind of work.
Skills are the brain. Skills are domain expertise encoded as markdown that Claude draws on automatically when the context calls for it. Claude doesn’t just connect to your data. It knows what to do with it. How to build comps. How to format a three-statement model. How to apply industry-standard color coding (blue for inputs, black for formulas, green for links). How to stress-test scenarios.
The mental model: Connectors bring data in. Skills know what to do with it. Plugins bundle both for your specific job.
The Excel workbooks the skills build
/comps [company] builds a comparable company analysis. Not a rough sketch. A formatted comp table with the right multiples, sourced data, and industry-standard layout. /dcf [company] generates a DCF valuation model with assumptions you can adjust. /lbo builds leveraged buyout models. /3-statement creates integrated financial statements.
The Excel workbooks it generates use live formulas with proper color coding: blue for inputs, black for formulas, green for links to other sheets. If you’ve spent any time in banking or FP&A, you know how much that matters. It’s the difference between a model someone can actually use and one that needs to be rebuilt.
The honest take
This is genuinely useful AND still early.
What works today: Google Workspace integration is solid. Asking Claude to search your Drive, read an email thread, and synthesize information across sources, that works well and saves real time. The payment connectors (Stripe, PayPal) are reliable for pulling transaction data. The Cowork finance plugins are impressive for structured workflows like comps and model updates.
What’s still clunky: The Excel sidebar is in research preview, and it shows. Complex multi-sheet workbooks can trip it up. The FactSet and MSCI connectors work, but getting the most out of them requires knowing what to ask for. They’re not going to replace an analyst who already knows their way around a terminal. And the cross-app workflow (Excel to PowerPoint) occasionally drops context on longer sessions.
What I’d set up first: Google Workspace. Highest-impact, lowest-effort. You’ll use it every day within the first week. Then Stripe or whichever payment processor you use. Then pick one plugin that matches your actual job and learn its slash commands.
Start small. Connect one thing. See what breaks.
Download the Claude Integrations Guide for Finance Pros (PDF)