In this week's @enopdf tasty treat I'm talking tables, tables, tables!
Parsing tables.
Copying tables.
Copying tables into Excel.
Copying tables into Markdown.
Annnnndd.......visualizing tables!
All of this is completely free and completely AI free. Try it out today!
When I took corporate tax in law school (one of my worst grades), I never imagined I'd one day be producing influencer content on the Canadian Budget Drop... but here I am.
In this week's @enopdf tasty treat I get to the bottom of the new entrepreneur capital gains credit in Budget 2026. The details are surprising!
I have no idea what our latest feature is for but it's cool so....whatever, we're dropping it
This is something that does not use any AI but never would have happened without AI - the cost/benefit equation never would have made sense.
Earlier this week I shared that we added table parsing in @enopdf. That made it possible to copy paste tables from PDFs into Excel with perfect formatting. It also made our AI search perform much better on table rich documents.
Since we now had the table data available, we thought, "what if you could just mess with this data without leaving Eno PDF?"
Thanks to the magic of AI, a few hours later we had a prototype where we could display a table as a chart. It was kinda neat so we ran with it and built out a robust implementation.
So today, you can take your boring tables and turn them into pretty graphs or pie charts with just a couple clicks. It's kind of remarkable how trends just pop out when you see them visually.
Give it a try - this is another completely free forever feature!
Everyone is trying to cram an AI chatbot into their app, but the real AI win might be more subtle.
Now that a small team can write hundreds of thousands of lines of code while they sleep at night, what is it possible to build that never would have made sense before?
This month at @enopdf, we set out to slay the beast that is tables in PDFs. Here's what we wanted to do:
- Copy and paste tables into Excel and have them layout correctly.
- Parse crappy scanned tables just a well as PDFs with digital data.
- Treat bracketed numbers as negative.
- Handle merged headers.
- UI to touch up the parsing when it's slightly off.
How did we do it? 14,000 sample pages, 6,000 labelled tables, and a TON of tokens.
This culminated in 50,000 lines of code that parse tables deterministically. NO LLM REQUIRED. This means it's super fast (120 pages/second on table heavy documents*), your documents stay completely local, and we're offering it completely free.
* That's single core performance. If you have a multi-core processor it's many times faster.
It's time for Eno Geoff to drop his latest tasty treat. Ctrl-F for those who don't have time to mess around.
It even works with grotty scanned pages. Bet your granny's PDF reader can't do this.