Hi everyone,

we’re currently facing a challenge in automating the analysis of larger PDF documents (60–80 pages) using GPT in Zapier.

👉 Our use case:
We want to check formal criteria in documents like real estate reports, such as:

• Is a cover page present?

• Is there a signature and date on the final page?

This works well when uploading the PDF directly into ChatGPT via the web app – the model can identify these elements with a good degree of accuracy.

However, when using Zapier, we’re running into two major issues:

1. The full PDF (especially longer files) often cannot be parsed or passed to GPT reliably.

2. Even when passing specific text excerpts (e.g., page 1 or the last page), GPT in Zapier tends to hallucinate results rather than strictly confirming whether a signature or cover page is truly present.

❓ What we’re looking for:

• What’s the best way to extract and structure PDF contents for reliable GPT analysis in Zapier, Make, or n8n?

• Which tools (PDF parsers, converters, etc.) are working well in your setup?

• Any best practices for feeding GPT only key pages (e.g., cover and signature pages)?

• How do you manage GPT’s context limits in these automation platforms?

We would really appreciate any technical advice, workflows, or tools you’ve found to solve this!

Thanks a lot in advance,
Paul --Details removed by moderator--
Sunside AI