How to auto-train documents using UiPath IXP for unstructured and complex data extraction

Help Document Understanding
1

Hi ,

I am using UiPath IXP for extracting data from unstructured and complex documents .

I want to understand how auto-training works in IXP:

  1. How can we enable auto-training for documents?
  2. Is there any specific activity required for auto-training in UiPath?

Please guide me on the correct approach and best practices to auto-train documents using IXP.

Thanks in advance for your help.

Regards
Sourav Bhunia

2

hi @sourav.bhunia

IXP Auto-Training (Generative Extraction):

  1. No specific UiPath activity needed - auto-training happens in IXP web interface
  2. Enable: Create Generative Extraction project → Upload 10-20 samples → Define schema → Write prompt → Auto-predicts
  3. Train: Validate/correct predictions → Refine prompt → Repeat (no ML training required)
  4. Use: Publish model → Call with IXP Extract activity in Studio

Key: Prompt engineering + human validation loop improves accuracy iteratively​

Best Practice: Start with diverse samples, aim for >80% confidence, use Orchestrator queues for production validation.

Regards!

3

Hi @sourav.bhunia

Pls check:

Happy Automation

4

Hi @sourav.bhunia

You will find some insights on your usecase,
Please find below youtube video to get more closer information about UiPath DU using UiPath IXP

5

@sourav.bhunia

There is no training or auto training involved as of now with UiPath IXP

extraction and all depends on the prompts given at document and field levels

cheers