Thank you

yes you probably could just search ctrl+f. I’m quite lazy so imagined that if I could upload the document (46 pages PDF of companies and their brands), I could then just chat with the bot/or perhaps take a picture of a fridge/products and it could quickly give me the answer.

not deliberately, however when run it did automatically identify a brand (Hellman’s) owned by a multinational NOT in the list (Unilever). I hadn’t included Unilever in the source doc as it is British/Dutch. So it must have used its own knowledge base.

Perhaps it interpreted my preference as wishing to avoid multinationals? This is what it output:

“Based on the list you provided and the document, here are the brands you should avoid if you are following the list:

1. Doritos: Owned by PepsiCo.
2. Tropicana: Owned by PepsiCo.
3. Mentos: Owned by Perfetti Van Melle, not listed in your document, but worth noting.
4. Hellmann’s: Owned by Unilever, not listed in your document, but worth noting.

The other brands you mentioned are not listed in the document as brands to avoid. However, it’s always good to double-check with the latest information or packaging details, as brand ownership can change.”

May help if I clarify. The document is ingested into the chat, so acts as a local knowledge base. When I put some random brands into the llm it uses the info from the doc to respond.

I probably could manually search the document (ctrl-f), just find it easier to use the llm approach, especially if inputting in an unstructured/messy way.

Sorry, I’m not familiar with the reference.

Important perspective. Admittedly I use AI all the time for work so using it in other areas has become second nature (for bettter or worse!)