Your analyst. Inside Excel.

Claude now works inside Excel as a side panel, right next to your model. Ask in plain English what's driving your returns or which assumptions are aggressive, and it answers with a link to the exact cell.

Stop trusting the model. Start interrogating it.

Most underwriting happens in a spreadsheet. The moment a number looks off, you're scrolling through tabs trying to work out which cell drives it. Claude for Excel sits beside the model and answers in plain English, every claim linked back to the exact cell it came from.

Most people accept the returns on someone else's spreadsheet at face value. Now you can ask which assumptions are aggressive.

Five prompts. Any model.

01

Trace the returns

Ask which three or four assumptions move the headline return the most, cited by cell, before trusting the number.

02

Stress the downside

Rerun the model with a tougher rent, void, and rate assumption and see which year, if any, turns negative.

03

Flag the aggressive assumptions

Go line by line and mark which inputs look optimistic against normal market ranges, with a more conservative figure suggested for each.

04

Find the break-even and compare deals

Get the exact tipping point on rent, occupancy, or exit yield, and compare two models side by side on the same terms.

The aggressive assumptions prompt

Go through every assumption in this model and flag which ones look optimistic against normal market ranges. For each, show the cell, the current value, and a more conservative figure you'd use instead.

What it still gets wrong.

i.

It can edit your model

Work on a copy for anything important and check every change before relying on it.

ii.

Check the cited cells

An answer is only as good as your verification of the cells it points to.

iii.

Benchmarks need local input

Supply a few real local ranges for rent growth, void, and exit yield, or treat the flags as prompts to check, not verdicts.

Next Playbook