Now assist doc intel extraction issue

[MDBilalI]

Hello All,

Actually, my scenario is when someone add a document in a record for example Incident then doc intel extracts the important points from that and provided in the additional comments.
For this scenario I created a Doc intel use case and it is doing the extraction then passing that to Now assist skill to format that extraction in a better form, but this is working only if the document has 5-6 pages, if pages like 10, 15 or more than that it is not working.
Do you have any idea like what could be issue here? In doc intel use case, I am using Now LLM.

Thanks,

Bilal

[boteeuwen]

This sounds more like a document size or model input limit than an issue with your extraction logic. With smaller files, Doc Intel can complete the extraction and pass the result to the Now Assist skill, but larger documents can hit limits in either Document Intelligence or the Now LLM.

One thing I would check first is whether Image Mode or table extraction is enabled in your Doc Intel use case, because those features have their own page limits. After that, check how much extracted text is being passed into the Now Assist skill. A document with 10 or 15 pages can produce a lot of text, and if the full extraction is sent in a single prompt, it can exceed the model context limit.

A good test would be to run the same document through Doc Intel and check whether the extraction itself completes. If it does, but the formatting step fails, then the problem is most likely the amount of text being sent to the Now LLM. In that case, splitting the extracted content into smaller sections before sending it to Now Assist would be worth trying.

If this helped, could you mark it as a solution or helpful?