Fabian Baier

245 posts

Fabian Baier banner
Fabian Baier

Fabian Baier

@baierSF

CEO & Founder https://t.co/hzkGEEsK81

San Francisco, CA เข้าร่วม Aralık 2013
320 กำลังติดตาม208 ผู้ติดตาม
Paul Graham
Paul Graham@paulg·
The reason AI coding works so well is that the source code of the median app was already slop before LLMs.
English
326
403
7.9K
501.2K
Fabian Baier
Fabian Baier@baierSF·
🤯 INSANE 24hrs: Wine with Sam Altman → Product Hunt launch → #3 spot with 265+ votes! But the REAL story? MCP servers just changed the game 🧵 We're literally giving AI "hands" - any AI can now talk to ANY tool/API Just built a Product Hunt MCP server in minutes. Our AI agents are now pulling live trending data through simple conversation 🔥 This is how we democratize AI. Every business can give their assistants superpowers to access their unique data. The future is NOW 🚀 Full demo: loom.com/share/3fac0a7b… #AI #ProductHunt #MCP #SamAltman
English
3
1
3
211
Fabian Baier
Fabian Baier@baierSF·
TGIF: it is 2025 and o3-mini's knowledge cutoff is still October, 2023 - what's up with that?! Did we really ran out of data throughout 2024 or do they no longer trust their own data after the "singularity" event because it could be hallucinated? lol
English
2
0
2
146
Fabian Baier
Fabian Baier@baierSF·
89% is inference. 11% is training. 100% is AI.
English
0
0
2
79
Fabian Baier
Fabian Baier@baierSF·
Open source is open source is open source - no, deepseek not spying on you or trying to infiltrate your thoughts more than an o4 would
English
0
0
1
90
Fabian Baier
Fabian Baier@baierSF·
So I no longer search on Google or bing. Instead, I find myself all day long in @PulzeAi Spaces talking to all AIs, searching & building.
English
0
0
1
43
Fabian Baier
Fabian Baier@baierSF·
AI training is a big business (exhausted the data for it need to drill or refine more) BUT AI inference is even bigger.
English
0
0
1
57
Fabian Baier
Fabian Baier@baierSF·
Our cutting-edge RAG Pipeline has set a new industry benchmark! We ingested over 50,000+ pages from the FinanceBench dataset (228 PDF files, each 150-250 pages long, ~525MB total) in just 1 1/2 hours. Our system outperformed existing benchmarks by a staggering 236% in the Shared Store configuration - currently one of the best results worldwide. Our pipeline is ideal for financial data, excelling at extracting and understanding large table data. This breakthrough demonstrates our position as an industry leader in financial data processing and RAG technology. #AIinFinance #RAGPipeline #FinanceBench
Fabian Baier tweet media
English
0
0
0
76
Fabian Baier
Fabian Baier@baierSF·
2025 will be all about agents. The first wave will be super simple but very powerful. Think of it as You provide the recipe and the AI automates it for you. Last year we automated model selection, this year we automate tool & agent selection.
English
0
0
0
54
Fabian Baier
Fabian Baier@baierSF·
Data is the new oil, but like fossil fuels, it's finite. We're burning through the internet's knowledge reserves, and now it's time to build agents that can refine and amplify what we have. #agentsnotdata #AI
English
0
0
0
49
Fabian Baier
Fabian Baier@baierSF·
🔍 So next time you interact with AI, remember: there’s a whole world of hidden thoughts behind the scenes! 🌟
English
1
0
0
38
Fabian Baier
Fabian Baier@baierSF·
🤔 What’s a latent space? Let’s break it down!
English
1
0
0
62