Michael Sakowski on Crunched: Excel-native AI analyst for finance power users that caught a £10M valuation error

Crunched is an Excel-native AI analyst built specifically for finance power users — investment bankers, private equity associates, and management consultants. Co-founder Michael Sakowski describes it as a side-panel chat inside Excel, similar to Cursor but for what he calls the world's most popular programming language. The target market is roughly 5 million users, the top 1% of Excel's 2 billion-strong base.

“Crunched is your Excel AI analyst built by and for power users. It's like this side panel chat in Excel, basically cursor for the world's most popular programming language. On a live deal for an associate at one of our private equity clients in London, Crunch's mistake detection system identified a mistake in the working capital that overvalued the deal by £10,000,000.”

Microsoft positioning

Sakowski is direct about the competitive dynamic: Microsoft is building Copilot for the full 2 billion-user base and is constrained by its competition with Google Sheets. Crunched is building for a narrower, higher-value slice where generic tooling falls short. The differentiation argument rests on depth of domain expertise — Sakowski and his co-founder each have more than 10,000 real Excel hours from prior roles at McKinsey and in finance, and the product is designed around workflows that analysts actually run: error detection, template augmentation, and model review, not just building basic analysis from scratch.

The data flywheel problem

The customer base creates an immediate tension: enterprise finance clients won't allow training on their data, which is live deal information rather than open-source code. Crunched's answer is that it does not train on customer data and cannot see what users prompt. The longer-term path is firm-level fine-tuning where large organizations do enough global modeling work to justify it, combined with forward-deployed customization — linking outputs into clients' specific LBO templates, formatting standards, and internal workflows.

Error detection in practice

The clearest proof point Sakowski offers is a working capital error caught by Crunched's detection system on a live transaction for a private equity client in London. The mistake overvalued the deal by £10 million. The deal itself involved a company described only as a $500 billion firm working on a $1.4 trillion transaction with roughly $20 billion in revenue — no names given.

The commercial logic is straightforward: if a single error catch saves £10 million, the ROI conversation with a finance team is easy. Sakowski notes the fundraise closed in record time, though no specific amount or lead investor is disclosed in the conversation.