Automating Debt Analysis with Claude for Excel

If you’re an investor analyzing a company with significant debt, you need to understand the debt waterfall.

Whether you’re a fixed income investor evaluating credit risk, or an equity investor trying to understand what sits above you in the capital structure… the debt waterfall is foundational.

But building one is painful.

You’re hunting through 10-Qs and 10-Ks. Cross-referencing footnotes. Manually pulling maturity dates, coupon rates, conversion terms, put options, call provisions. Then organizing it all into something you can actually model off of.

For a company with multiple tranches of debt, this can take hours. Sometimes an entire day.

This is where Claude for Excel changes the game.

Anthropic recently launched Claude for Excel, a plugin that lets you run AI prompts directly inside your spreadsheet. I’ve been testing it for finance workflows and the use case that immediately stood out to me is debt extraction.

The idea is this:

1. Feed Claude for Excel the raw 10-Q or 10-K filing
2. Run one comprehensive prompt that extracts all debt instruments, terms, covenants, and qualitative footnotes
3. Get back a structured Excel workbook you can model off of immediately

What used to take hours of manual scrubbing now takes minutes.

I’m going to walk you through this workflow using Strategy’s (formerly MicroStrategy) Q3 2025 10-Q as an example.

Strategy is a good test case because their capital structure is complex. Six convertible note series. Over $8 billion in principal. Each tranche has different maturity dates, conversion prices, holder put options, and issuer call conditions. The kind of debt stack that would normally take significant time to properly map out.

Let’s see how Claude for Excel handles it.

The Result

Full output here: MSTR_Debt_Waterfall.xlsx​

Claude for Excel produced a four-tab workbook in one pass:

Tab 1: Debt Master Table: Every debt instrument mapped across 38 columns. Principal amounts, maturity dates, coupon rates, conversion prices, put dates, call conditions, settlement methods, ranking in the capital structure. All sourced and cited to the specific page in the 10-Q.

Tab 2: Debt Events Ledger: A chronological log of every debt event. Issuances, redemptions, conversions. Cash paid or received. Accounting impact. This is the audit trail you need to understand how the debt stack evolved.

Tab 3: Qualitative Footnotes: The nuanced language that matters. Fundamental change definitions. Anti-dilution provisions. Events of default triggers. The stuff that’s easy to miss when you’re manually scanning footnotes.

Tab 4: Open Items: This is the part that impressed me most. Claude flagged what’s NOT in the filing. Missing covenant details. Undisclosed collateral descriptions. Cross-default provisions that would require pulling the original indentures.

In other words, it’s not just extracting data. It’s telling you where the gaps are in your diligence.

The Bigger Picture

This workflow fundamentally changes how you can approach credit analysis.

Think about the typical process. You pull the 10-Q. You ctrl+F for “debt” and “notes payable.” You manually build a table in Excel. You cross-reference three different footnotes to figure out the put date on one tranche. You miss the conversion rate adjustment language buried on page 47.

Now you can upload the filing, run one prompt, and get a structured output that would have taken hours to build manually.

A few things I’d highlight:

• The output is model-ready. You can immediately start building a maturity schedule, modeling conversion scenarios, or stress-testing liquidity.
• The citations matter. Every data point is traced back to the source. You’re not blindly trusting AI output. You can verify.
• The open items tab is your research checklist. It tells you exactly what to pull from EDGAR if you need to go deeper (indentures, 8-Ks, security agreements).
• This scales. Run the same prompt structure across the debt stack of five different companies you’re comparing. What used to be a full day of work becomes an hour.

Limitations to keep in mind

Claude is pulling from what’s disclosed in the filing. If the company buries something in an exhibit or references an indenture without summarizing the key terms, you’ll still need to pull the primary source. The open items tab helps here, but it’s not a substitute for reading the docs on material positions.

Also, for the most complex structures (secured credit facilities with 200-page credit agreements), you may need to feed Claude the actual agreement rather than relying on the 10-Q summary.

Bottom line

If you’re doing any kind of credit work or analyzing companies with meaningful leverage, this workflow is worth adding to your toolkit.

Try it on the next 10-Q you’re reading. I’d recommend starting with a company that has 3-5 debt tranches to get a feel for how Claude structures the output.

Hit reply if you have questions on the prompt structure or want to see this applied to a different sector.