💙 If there were an #AwesomeAdmin manifesto, it might sound like this:
We listen before we build.
We turn messy processes into better ways of working.
We make data reliable, experiences useful, and change less daunting.
We know when to put AI to work and when to trust our own judgment.
We make the technology work for the people who depend on it.
@kr0der For ads duh! How are you gonna buy that thing that ChatGPT told you you’re gonna need that you actually don’t need if you don’t know where to find it?
@mtteo22@Benioff@SlackHQ@salesforce@AnthropicAI At no point have I ever heard him fund communists… Socialists maybe but not communists.
He earns like 5% of Hawaii…. Don’t think he’s really trying to help bring the communists to power.
@flyek91@testingcatalog Depends on the case. I’m so so I need to create a recent text based and don’t have any kind. It’s not enough to justify a full project for them since they’re gonna be rarely used and have no logic attached to them, all it is a spot to store a little data
Dear Lord, you've taken my favorite singer, Dolly Parton, my favorite comedian, Catherine O'Hara, and my favorite singer, Bonnie Tyler. I just wanted to let you know my favorite politician is Donald Trump.
Why the fuck does AI psychosis always manifest as spamming useless network graphs and tons of animated elements?
It’s absurd how common this very specific motif is
A small AI builder cheatsheet I wish I had when I started.
Embeddings → Sentence Transformers
Vector search → FAISS
Vector database → Qdrant
LLM API → OpenRouter
Local LLMs → Ollama
Structured output → Pydantic
API backend → FastAPI
Agent workflows → LangGraph
Local speech-to-text → Whisper
Local text-to-speech → Kokoro
Model fine-tuning → PEFT + LoRA
Model hub → Hugging Face
Experiment tracking → MLflow
LLM evaluation → Ragas
LLM observability → Langfuse
AI app UI → Gradio
You don’t need all of them.
Pick the boring tool that solves the problem.
@AbeBurnett@kelvinbuildss Interesting pattern here. Are you defining? The actual technical implementation is part of your specs or just recommended approaches? Is Grok/deepseek then making the appropriate technical decision with Luna making the actual files/edits?
Trending repository of the day 📈
prime-agent
A self-improving RLM agent for coding workflows and long-running autonomous tasks.
Last 24h: 2,293 ⭐
Total: 7,568 ⭐️
https://t.co/FhaGrDOgfd
@UzerKayat@perplexity_ai For the $20 you get
1.) Perplexity Search and choice of models
2.) Comet Browser
3. Connectors and Workflows
Claude and ChatGPT publish the models which even Perplexity runs on. Its like a gasoline company making a car, you wouldnt then go to Honda and demand free gas