Claude Feldges
@ClaudeFeldges
Followers
50
Following
569
Media
0
Statuses
143
Passionate about Data Science, ML, AI, NLP. Physicist with 15 years of experience in Private Equity, from Risk Management to Data Science.
Joined August 2021
Great news: Switzerland‘s best universities have developed a fully open LLM, Apertus! Excited to see the adoption by Swiss administration and Swiss corporates. I predict this will be huge 😀 Link from press release from ETHZ in comments.
1
1
3
🔗 https://t.co/zTdS9Ij0u6 Key insights from building internal chatbots: - Model-agnostic architecture saves time & money - Avoid vendor lock-in from day one - Future-proof your AI investments 🚀
aipe.tech
Consulting firm turning AI potential into working business solutions
0
0
0
For companies, switching AI models can take weeks ⏰ and significant budget 💰 - even for simple upgrades. ⬆️ Smart architectural decisions early on 🏗️ could prevent this trap. My article below 👇
1
0
0
AI Vendor: "We'll replace your workforce by Q3!" Smart Company: "Let's automate one task first." Guess which delivers ROI? Real AI wins happen through small steps, not grand promises. Why starting small beats snake oil:
medium.com
I recently spoke with a tech leader at a major company who shared something that’s been bothering me. Despite being sharp technically, he’d…
0
0
1
Why does AI diffuse slowly in corporate environment? I wrote about it in my latest article:
0
0
0
Reading docs on smolagents from @huggingface . Great stuff! But somehow it reminds me of DSPy. I feel I work with agents since early 2023, when I started using DSPy. It just did not have a fancy name back then… 🤔 Thanks @lateinteraction for the innovative work!
0
0
5
I have tested @browser_use - the framework used by @ManusAI - on a complicated web extraction 🔍. While there are occasional hiccups (the bot can get stuck reading policies) - overall it works really well! 💻 I have reported my testing in an article 📝 - Link in comments below 👇
1
0
0
I just published DeepSeek’s R1 is Great. Here’s the Catch.
link.medium.com
As many have written recently, the Chinese company DeepSeek deployed two very competitive language models, V3, and R1. The latter gained…
0
0
0
First time using STORM+DSPy+FastHTML in production - feedback welcome!
0
0
0
Try it: - https://t.co/DaEMDPR7QS (Google auth sign-in) - Enter company + context ("Tesla, an EV manufacturer" works better than "Tesla") - Get report in 2-3 mins Five free reports to start.
1
0
0
Under the hood: - Research phase: 4 experts scanning 100+ sources in parallel - Writing phase: 10 concurrent writers for fast execution - Data: Real-time Bing Search integration
1
0
0
Tech stack: Stanford STORM for report generation with parallel agents + FastHTML for UI Result: Full investment report in 2-3 mins Demo:
1
0
0
Built my first real-world DSPy+STORM+FastHTML app: an AI investment report generator! 🚀
1
0
0
Need more reports? More features? Customization? Contact me via DM. Do you like it? Then share with fellow analysts!
0
0
0
The report contains key insights about the company, powered by 100+ analyzed sources. Five free reports to start.
1
0
0
How to try it: - Visit https://t.co/DaEMDPR7QS - Sign in with your Google account - Accept terms - Enter company name - Wait 2-3 mins - Get your report Tip: Add context ("Tesla, an EV manufacturer", not just "Tesla") to guide the app.
2
0
0
Result: Professional investment reports in 2-3 minutes.
1
0
0