I am currently facing an issue with anAI model in Power Automate (AI Builder)that is designed to extract data from PDF documents. While the model consistently shows a highaccuracy score of 99%, itfails to extract several expected fields correctlyduring real-world execution. Please feel free to guide me if there any workarounds. Thank you.
The accuracy score in AI Builder is based on your training data, not live documents. So if your production PDFs differ even slightly — in layout, font, spacing, or field presence — the model may struggle despite a high score.
I'd include real-world PDFs that reflect the variety you’re seeing in production.
Was this reply helpful?YesNo
Under review
Thank you for your reply! To ensure a great experience for everyone, your content is awaiting approval by our Community Managers. Please check back later.
Share
Report
All responses (
Answers (
710
