[ivanvanderbyl]

Congrats on the launch!

I’m curious to hear more about your pivot from AI workflow builder to document parsing. I can see correlations there, but that original idea seems like a much larger opportunity than parsing PDFs to tables in what is an already very crowded space. What verticals did you find have this problem specifically that gave you enough conviction to pivot?

[maxmaio]

We saw initial traction with real estate firms extracting property data like rent rolls. But we've also seen traction in other verticals like accounting and intake forms. The original idea was very ambitious and when talking to potential customers they all seemed to be happy with the existing players.

[smt88]

How do you guarantee that nothing in an extracted rent roll is hallucinated?

[themanmaran]

The same way you guarantee that a person manually typing the data never makes a mistake.

[smt88]

Humans in this space tend to make mistakes like, "Added rent as per-square-foot instead of absolute value," or, "Missed a rent escalation for year 3."

These tend to be easy to catch, even for the same person who's reviewing the data. They would see that rent steps looked strange (Y2 and Y4, but not Y3) or there was an order-of-magnitude difference in rent from one month to another.

AI can do something like invent reasonable-looking rent steps. They're designed to create output that seems reasonable, even if it's completely made up.

When humans are wrong, they tend to misread what's there, which is much less insidious than inventing something.

And if you have a human reviewer for all the work this AI does, what's the point of the AI in the first place? The human has become the source of truth either way.