mrbishalsaha

For me, I believe building a small MVP myself, scaling it and then building a proper v1 with a better foundation and team.

Repeat every 3 years minimum (if viable after PMF), learn design principles, core USP offerings, remove spaghetti code and building documentation.

But how? I didn't find any research paper or specifications in their developer pages.

How can an LLM be trained on realtime data? If not than how does the pplx-70b-online be online?

In my understanding, it's just doing a function call for all the queries using Bing and replying with citations. Nothing new even, Mistral and ChatGPT-4 does it. Also if vector embeddings are being used how can it be scalable?

State Bank of India is obviously the biggest with 50% or close rest are mostly private banks.

Nothing huge technological change in this side yet.

No crypto. I bet on crypto once as a dev and I see no future.

As for 'mortgage broker', I am building one actually in India

Honestly, I am trying to gather more user feedback. I've seen traffic of 2.5k so far, but most visitors didn't sign up, start a conversation, or do anything; they just left. I can't learn anything if people don't even complain.

I understand that uploading private information to an unknown site, even for a trial, is very difficult. So, instead of asking for bank statements, I am asking users to start with any PDF. The reason I'm asking for this is because without context, it's just another ChatGPT.

I want to build a local LLM to operate on devices, but currently, most LLMs don't run well. I'm hoping this will change with WWDC when Apple announces new iOS updates and new AI features.

I am desperate to get some actionable feedback that I can work on to improve, or I might just have to abandon this project. That's the reason I'm reaching out. Sorry for spamming.