Hit a few speed bumps tho, including my attempt to integrate an external agent. they ended up talking to each other instead of working together to solve the problem. Not what I wanted but it was quite entertaining.

Got distracted with side projects, but ended up improving the call assistant dramatically! The flow is much better and it transcribes and summarizes the conversation. Quite a few use cases for this imo.

I spent the last few days making a call assistant. Most of that time was used to get the proper transcriptions saved while my assistant answers in real-time. Can't wait to use this to screen all the scam calls I get!.

Wasn't sure what to do beyond image/video generation so I pivoted to a VTuber avatar creator and simulator where you can export your creation. I left the original functionality and users can switch between the versions with a toggle on their profile page.

Spent most of my week exploring different AI Image/Video generation models. I really need to improve my prompting and creation flow because the results have been creepy to say the least. I'm still working on the UI/UX but it's viewable for now at least.

I need to reach out to some designers to improve the look and feel, definitely somewhere I'm lacking right now. Feel free to share any suggestions you have!.

Excited to share the progress on my first ever Flutter/Android app, MikMok! I spent more time than I should've on the backend but I think that'll be a good thing when I fully flesh out the features next week.

This week I was able to spin up both a Flutter and a Swift app in a few hours with Firebase authentication and storage. The video playback was a little hard to do but once I got that part working, being able to recreate an application similar to TikTok was a breeze.

Now that I'm more familiar with Supabase, I've been adding in some edge functions so an Agent can use them as tools to automate (almost) everything. I'm using LangSmith to measure the performance but still improving it. I think 88% reliability isn't so bad of a start though.

Made quite a bit of progress in the past few days. I'm still ramping up but I feel the progress! The demo even required two full videos this time since there were so many features. (Or maybe my recording script needs work)

Turns out my problems with Supabase were user issue, oof. I learned a lot after discovering the Authentication tab. Anyways, here's a much better demo of the basic functionality once you're authenticated.

Any supabase experts? I heard it was easy but I can't even get the authentication to work. Why is there infinite recursion in reference to user?!! Banging my head over this right now.

After some tinkering around with some scripts exploring RAG, I've managed to improve my existing functionality to reflect desired context and added another feature to reply for you when you're offline. And this is just the tip of the iceberg!

Forgive the UI but I thought it was cool to integrate Retrieval Augmented Generation (or RAG) to provide context to the model from messages previously sent through the application before providing an answer 😎

After a few more days of soaking up tons of information on web development and working alongside an AI companion, I've been able to put together a (mostly) working messaging application. Got tangled up in some wrong commits in the demo though 😬

video of auth flow as well:

Just wanted to share what I've been working on the past 2 days. It may not seem like much, but I started with zero knowledge of web development and hardly any experience using LLMs/AI. Looking forward to what I can accomplish going forward!

Is this thing on?.