Five months ago, in March, I stumbled across something that would eventually change the direction of my work. A programmer from Pakistan had just released a tool called excel-mcp-server.

For context, MCP (Model Context Protocol) is a framework introduced by Anthropic-the company behind Claude-at the end of last year. It quickly became a game-changer, standardizing how large language models connect with external tools like Google Drive, YouTube, Reddit, GitHub, and countless others. At the time, MCP adoption was exploding.

After installing excel-mcp-server, suddenly my Claude instance could talk directly to Excel.

The first thing I did-like any ex-investment banker would-was download a company's financial statements, feed them into Claude, and ask it to build me a valuation model.

The result? Pretty rough. The model was clumsy, but it existed. And that single attempt planted a thought in my mind: this could be a product.

But I set it aside and went back to my day job.

The Reddit Spark

One month passed. Then two. Every now and then I'd search to see if anyone had cracked the "AI builds financial models in Excel" problem. Nothing-just people using ChatGPT in a consulting style, never integrated with Excel.

I couldn't believe no one had built it yet. It nagged at me constantly.

Then one Saturday in mid-May, I came across a Reddit post titled:
"An AI that can build Financial Models in Excel"

Finally, someone had done it. And it wasn't me.

I tested the product immediately. It was buggy, slow, and limited-but it worked. And in that moment, I decided: no more sitting on the idea. I would build my own version. I made a commitment to launch what would become iLoveDCF.

What iLoveDCF Does

iLoveDCF is an Excel Discounted Cash Flow (DCF) model builder for American and Brazilian companies.

Here's how it works:

You drop in the company's financial statements (PDF).
In about 3-6 minutes, you get a full DCF model: historicals, projections, and valuation.

It's still an MVP:

Only one PDF at a time.
Annual statements only.
Assumptions are fixed, based purely on historicals.
Equity is valued by book value, not market cap.

And yet, if I'd had this back in my analyst days, it would have saved me hours. I would have been a user.

The Hard Part: Distribution

Building the product was only step one. Now comes the harder challenge: getting it into people's hands.

Distribution, in my view, is even more important than product. A great product with weak distribution dies. A decent product with strong distribution can thrive.

Here's my current playbook:

LinkedIn.
Reddit (tricky, given their stance on promotion).
Personal contacts.
Maybe some cold outreach to boutique M&A firms and funds. But I'm not in the mood to do that, honestly.
Commenting on posts from people building similar products.

Mundane Problems, Big Opportunities

Recently, I came across a solo founder making $40k/month just by converting bank statements from PDF to Excel. His story is here: link.

And yes, the name iLoveDCF was 100% inspired by iLovePDF-the PDF-splitting/merging site that's one of the top 50 most-visited websites in the world. Who knows? Maybe a little halo effect will rub off.

The truth is, the world is full of small, "boring" problems that can become massive opportunities. AI makes it easier than ever to explore them.

If you liked iLoveDCF, or you want a bespoke solution for you, please write to me.

Lucas

[email protected]