used to be a certified matcha hater until one specific spot completely converted me 💀 so i tried literally every matcha bar in Indiranagar to give y'all the ultimate ranking. fyi I'm not a sweet tooth at all, so this is 100% based on my own palate. let's get into it 👇
why is all the techbro advice on twitter so cultish and believing in a higher purpose, why are they so needy to create struggle, where is the god complex coming from
you are not unique, you are the sum of your surroundings and we can all enjoy and build things together 👨❤️💋👨
folks of bangalore, i'm selling two Vir Das: Sounds of India tickets for 25th Oct, 8 PM at buying price. the seats are centrally located (M row). dm me if you want both/ would like to sit beside me.
@thevirdas
the way every meet with my manager makes me feel optimistic about the future of the product whereas one meet with the TL is enough to make me spiral about this piece of shit godforsaken app
3 months ago, our research team decided to lock-in and architect a new memory system for AI agents, cause nothing out there that worked for our use cases
We call it Prem Cortex 🧠
Today, we're open-sourcing it, so you can give your AI agents memory that f̶o̶r̶g̶e̶t̶s̶ scales!
> be me
> mom finds my instagram
> sends me two mildly inappropriate pics as proof
> gaslight her
> ask what's wrong with that
> confuse conservative indian
> laugh uncontrollably cause what are they gonna do about it from thousands of miles away
Saw a college senior’s Instagram story and had to double check like, seriously??? Since when does it matter if your delivery guy eats non-veg? Is he delivering food or his diet plan? 😭
Copilot for programmers is almost hitting convergence, however copilot for data analysts is still in emergence. Introducing @premai_io PremSQL Agents, AgentServer and Playground, where you can use our prebuilt agents or make your own for autonomous data analysis.
ps: this demo that you see runs fully local using Prem-1B-SQL for text to SQL and LLama 3.2 for analysis and plotting on Apple MLX . Even ollama just takes ~ 5GB of vram running the two at the same time on gpu machines. With our latest release:
> connect to data sources (postgres, mysql, sqlite, csvs, kaggle datasets etc) and query them using natural language.
> do analysis over tables and plot charts
> You can even use it with different programming languages with our Agent API.
> we support PremAI, HuggingFace, Ollama, Apple MLX, OpenAI as our model connector
> you can train your own text to SQL models or other workers with our datasets and tuner
> evaluate and compare models
And do lots of customization, like add your own database connection or custom inference engine etc. Check out the documentation and GitHub repo.
Folks on my tl, gathering feedback for @foldmoney_ mobile app, could you please reply/DM with
- What you love most about Fold
- Areas needing immediate improvement
- Critical missing features you'd like to see
Fold goodies are up for grabs for quality feedback.