Adding documents to AIP Agent

[kelkarshriram]

I m new to AIP agents and I created a simple AIP agent that uses PDF files but I am facing 2 issues.

1. It’s not learning, many of our products have common parts but it’s not recognizing that they are common though they have exact same names? How can I make it learn?
2. where can I find guided easy to follow training ? I didn’t find it on learn.palantir.com
3. how do I provide a media set as input to the agent, it asks me select files from the media set but if I later add files to media set, they are not auto captured or used.

[arukavina]

You can always parse the content of the media set, running an automation that executes a chunking pipeline to run embedding from the content.

Then, in Agent Studio, you select the ontology semantic search tool:

Pointing to the Object Type that has the embedding,

Alternative you can use context tool for retrieval:

Uploading a PDF.

For guides, please check these out:

• Semantic search • Use Palantir-provided models to create a semantic search workflow • Palantir
• Semantic search • Document processing • Palantir
• Semantic search • Ontology augmented generation • Palantir

Video: Building with Palantir AIP: Semantic Search

Blog: Building with Palantir AIP: Semantic Search | by Palantir | Palantir Blog

Hope this helps, please let me know.