Ai models

Help Agent Builder
1

Which AI model is best suited for large-scale data extraction from Excel files, and what factors should guide that choice?

2

@Benhamou_Hassan

For extraction you dont need an AI model

but if you need to analyze it or get some data depending on what you want to do you need to choose the models

for example if you have some old data in it which serves as reference for new data then adding it as context makes sense ..

cheers

3

I’d usually go with a strong general LLM like GPT 4o if I need to extract or clean messy Excel data because it can understand structure and reshape it pretty well, especially when paired with Python tools.

4

thank you for ur comment @Anil_G
I need an intelligent extraction from an Excel file that has no fixed or consistent format. The model must understand the content semantically and extract the relevant information based on meaning not column position or predefined structure

5

Hi @Benhamou_Hassan

I agree with @Anil_G in saying that for Excel extraction you shouldn’t really need an AI model.

However, if you’re really dealing with LARGE data sets (1 000 000+ records etc.) I would go with Claude 4.5 Opus or GPT-5+ for sentiment analysis.

I know Claude 4.5 Opus is no available by default in UiPath just yet, but there are other integration services that can consume it.

Good luck​:slightly_smiling_face:

6

@Benhamou_Hassan

To choose a specific model, you need ask few question:

Is your excel data have sensitive data? If yes, research on which model(s) are compliant
Is your input file have highly unstructured or long data or short and easy? If it’s ling and highly unstructured then you can check token limits and models which are good at processing long and messy data.

I also has a similar scenario but the data was plain text, large and messy, but my data sometimes had sensitive data, so I did choose GPT 4.5 model which was good for my scenario.

I did observed few things when testing different model (if it help you to narrow your search):
Claude - very good at processing large and messy data but (as per research) it was not compliant
GPT - older version (GPT 4o was compliant but has lower token limits and could not handle complex data), new versions (GPT 4.5, 5) were compliant for our requirement and also good at handling large and complex data)
Gemini - Balanced compared to above.

Hope it helps you to understand and choose further.

7

@Benhamou_Hassan

So basically what you need is a context and summarize with citations…so for that generally from available ones gpt 5.4 or claude haiku should be good

cheers

8

thank uuu all for ur comments .. appreciated

cheers

9

@Benhamou_Hassan If your Excel files are clean and structured (fixed columns, proper tables), honestly you don’t need any AI model. UiPath’s Excel activities + DataTable/LINQ will handle large-scale extraction much faster and cheaper.

AI models make sense only when things are messy—like:

  • different formats across files

  • missing or inconsistent headers

  • need to “understand” or map data instead of just read it

In those cases:

  • GPT-type models can help with mapping, transformation, or logic

  • UiPath Document Understanding can help if Excel is part of mixed documents

  • Custom ML if you have very specific patterns to detect