<https://johnoreilly.dev/posts/gemini-kotlin-multi...

# multiplatform

[Joel Denke]

https://johnoreilly.dev/posts/gemini-kotlin-multiplatform/
Anyone tried this but with some more customizations? I think about using embedded model and provide my own text it can search. I also checked that there is some Firebase extensions, where I think Gemini can search for content in GCM/Firebase directly, but I am not sure. Not many samples and I am quite new to practical level of LLMs. I would like to be able to ask anything from my own data, and not pre-trained data from Google, other than how it learned how to fetch data from input. Like given this database, search anything in database related to a users multi modal prompts.
Maybe there is even KMP libs doing exactly this, not sure. Any pointers in this topic?

[Colton Idle]

I'm going to give this a shot later this week.

[Joel Denke]

@Colton Idle Cool, if you feel for it, please slide in a DM or tell me here how it goes 🙂 I am soon need to implement AI, but want to use anything existing LLM.
In my case I kind of want a database indexed to LLM power and let users ask AI anything from the database indexed data, in combination with its existing models.
Hard to explain 😄 All cases I found so far is to basic and very hard follow both OpenAI and Gemini APIs when I am not a professional data scientist or such 😄

[Joel Denke]

Just want to say, here is my content, here is my input field, give me your best 😄