Vertex AI search conversation

[Manikantah]

I have no idea how to get recommendation of images with text and only text responses form DFlowCX integrated with Vertex AI search and conversation. please can anyone help me get there? I have tried to create all type of datasets but nothing worked.

is there anyway other than uploading each photo manually to the cloud and linking it to the datastore.

if anyone is working on similar thingy please DM me.

[SaurabhPandey]

@Manikantah Can you explain a little bit more on your use case and what you are trying to do with Vertex Search and Conversation?

[Manikantah]

we are trying to build a chatbot for a pharmaceutical company.Our requirement is RAG-like thing with Rich text response of the drugs within in the company and nothing beyond.This is what a sample conversation should look like BTW this is a user and pharmacist related chat bot.
User:Hi, i am feeling fatigue can you recommend me a medicine?

AI: Sure can you describe any another symptoms

User: Yes and also headache,sleeplessness

AI: Cool,i suspect you have Chronic Depression.I recommend you to contact a health professional.
User: Okay, can you recommend any drugs for time being?

AI: Sure here are my recommendations(Recommended use after consulting of professional advice only)
<Img1> with text(name,cost)
<Img2>with text(name,cost)
<Img3>with text(name,cost)

Thank you............... for considering

[SaurabhPandey]

Thank you for sharing the use case with me. I feel like in your case you might have to chain few things together to get images as part of the chat response. If you have a product catalog you might use multimodal embeddings to find the right text+images combination based on the input from the chat. As I said typically there are multiple way to solve a problem. One of the product in Google Cloud is  Vector Search that you can leverage for this use case https://cloud.google.com/vertex-ai/docs/vector-search/overview

Hope it helps!