The Challenge: Thousands of PDFs, No Backups

This morning started with an unpleasant surprise: a server malfunction had wiped out several backups and archives.
For one of our clients, this meant thousands of invoices and payment orders were suddenly available only as PDF files
and needed to be re-imported into the company’s ERP system.

Normally, this kind of task takes days of manual data entry, prone to errors and frustration.
I decided it was time to try something different and smarter.

The Solution: FileMaker 22 + OpenAI API

I turned to FileMaker 22 (latest release) with FileMaker Server and began experimenting with
the OpenAI API. Instead of relying on FileMaker’s native AI script steps, I chose a more direct approach.
I imported the raw text from each PDF and sent it straight to the API for analysis.

This gave me full control: the AI could see the original content of the PDFs, analyze it line by line,
and return structured data perfectly suited for the ERP interface.
The process avoided shortcuts and made the extraction more accurate.

The Results: Smarter Than Manual Entry

After just a few hours of development and testing, the outcome was astonishing.
FileMaker and AI worked hand in hand to not only parse the PDFs and extract key fields,
but also to provide richer and more accurate data than human operators normally entered.

One Click, Full Data

All I had to do was point to a PDF, and within seconds, the ERP fields linked to that document were populated:
clean, complete, and often enhanced with additional context that would have been lost during manual input.

The accuracy was higher, the data was more consistent, and the time savings were dramatic.
What used to take days could now be done in hours.

Why Not Use FileMaker’s Built-In AI Steps?

Some might wonder why I didn’t use FileMaker’s built-in AI script steps.
The reason is simple: I needed to control the exact content being analyzed.
By sending the raw PDF text directly to the OpenAI API, I could ensure that nothing was lost in translation
and that every piece of useful information was available for extraction.

This approach also allowed me to design a flexible workflow: the same method can be reused not just for invoices,
but also for contracts, purchase orders, HR documents, or any business PDF archive.

Business Value: Beyond Automation

The benefits of FileMaker 22 AI PDF Data Extraction go far beyond speed.
For businesses, it means fewer errors, faster reporting, and the ability to enrich ERP data with insights that were
previously overlooked. Operators are freed from repetitive tasks and can focus on higher-value work.

For managers, it means a more reliable data pipeline. For developers, it means building solutions that make
entire teams more productive. And for companies, it means a clear return on investment in both time and money.

Reflections on the Future of Work

This experiment showed me something very clear:
a few hours of work from an experienced developer can automate
up to 90% of the daily tasks performed by employees without technical skills.

The future where AI replaces many of our routine jobs is not science fiction—it’s already here.
But instead of fearing it, we need to understand it, learn it, and be the ones guiding it.
Because while AI can execute brilliantly, it still needs someone to set the rules…
at least until the day it learns to validate itself. By then, I’ll probably be in the mountains, herding goats.