I built too many features into my PDF app. So I cut most of them.

Title:

I built too many features into my PDF app. So I cut most of them.

Body:

I’ve been building a PDF tool called PDflow for a while.

It started with a very simple idea: PDF → Excel.

Then I kept adding things.

Merge, split, compress, watermark, templates, OCR, different conversion modes… At some point, I had built quite a lot.

And weirdly, the more complete the product became, the less sure I was about it.

Not because it didn’t work.

Because I started asking myself:

Why would anyone use this instead of Adobe, WPS, Smallpdf, iLovePDF, or one of the hundred PDF tools already out there?

I didn’t have a convincing answer.

“Free” isn’t a moat.
“Local processing” is useful, but probably not enough.
And trying to win by adding more PDF features felt like a race I was never going to win.

The thing that changed my mind came from testing harder PDF-to-Excel files.

A lot of them technically converted fine. The Excel opened. Nothing crashed.

But then you looked closer.

A row was shifted.
Wrapped text became two rows.
A table broke across pages.
A SKU matched the wrong quantity.

And the worst part was: the file still looked successful.

A user recently left a comment saying:

“Having to hunt for silent layout errors is what ruins the whole experience.”

That sentence really stuck with me.

Because maybe the real problem isn’t:

“Can this PDF be converted?”

Maybe it’s:

“Can I trust the result without checking everything again?”

So I decided to cut most of PDflow.

For now, I’m hiding merge, split, compression, watermarking, PDF-to-Word, PDF-to-PPT and the rest of the toolbox.

I want to focus on one workflow:

PDF → extract → detect suspicious data → review against the source → fix → export

If something looks wrong, PDflow should tell you where to look, and ideally show exactly where that value came from in the original PDF.

I’m also rebuilding part of the architecture, because the first version mixed UI, worker state, recovery logic and export logic too tightly. I spent way too much time fixing one thing and breaking another.

This time I’m trying to be much stricter.

No more features just because they sound useful.

First I want to answer one question:

Can PDflow make difficult PDF-to-Excel work meaningfully easier to verify?

If not, I’d rather find out now.

If you work with PDFs that “convert successfully” but still force you to manually check everything afterward, I’d genuinely like to hear what usually goes wrong.