Johan (@johanrin) header
Johan (@johanrin) profile picture

Architecting the future with AI & Cloud | Microsoft Foundry | Strategic insights from multiple cloud journeys

Paris
Joined May 2010
Photo by Johan (@johanrin)Photo by Johan (@johanrin)Photo by Johan (@johanrin)Photo by Johan (@johanrin)Photo by Johan (@johanrin)Photo by Johan (@johanrin)Photo by Johan (@johanrin)Photo by Johan (@johanrin)Photo by Johan (@johanrin)Photo by Johan (@johanrin)
Does no-code still exist? I mean, who wants to no-code when you can vibe code?
1
141
The AI industry doesn’t need Netflix. It is Netflix. New drama every week. 🍿
We’ve identified industrial-scale distillation attacks on our models by DeepSeek, Moonshot AI, and MiniMax. These labs created over 24,000 fraudulent accounts and generated over 16 million exchanges with Claude, extracting its capabilities to train and improve their own models.
1
203
In the meantime, your clients think AI = automation
Hype train don’t care about ROI
1
186
Me: WHY does my plan usage hit the limit so fast Also me: All in for Opus 4.6 🚀
2
165
Good news: Microsoft Agent Framework just hit Release Candidate. API surface stable. All v1.0 features done. Works with Microsoft Foundry or any model/AI service you want. Single assistant or a full team of specialized agents, same foundation. The fence for building agents for real just disappeared. devblogs.microsoft.com/found…
98
Hot take: your AI not one-shotting your code isn’t an AI problem. You never one-shotted it either.
1
3
121
Lost internet while using AI. Switched to my phone hotspot. I had time to make coffee between each token ☕️
114
Here comes a new challenger! 🥊
This is Claude Sonnet 4.6: our most capable Sonnet model yet. It’s a full upgrade across coding, computer use, long-context reasoning, agent planning, knowledge work, and design. It also features a 1M token context window in beta.
1
108
LLM routing using the right model.
2
116
You're using Claude Opus 4.6 for everything. Then you see the bill. 💀 What if the system could pick the right model for each prompt, automatically? That's exactly what Azure Model Router does. I wrote about it 👇 johanrin.com/posts/model-rou…
2
2
137
Unpopular opinion: Your feed is only as good as who you follow. I want more developers, AI nerds, and cloud architects in my timeline. If that’s you: 1. Like this 2. Drop what you’re building 3. Follow back Let’s grow together.
3
2
125
What if the LLM debate is the wrong debate? Open source might be the actual game changer here. Thoughts?
61
My feed right now: - Coding is almost solved. - Yay! I just crossed 500 followers! Half of tech X is having an existential crisis. The other half just discovered confetti emojis.
71
developers in 2020: googles "how to center a div" copies from Stack Overflow developers in 2025: asks Claude Opus 4.6 "how to center a div" Opus 4.6 googles it copies from Stack Overflow $20/month well spent
2
131
My current feed: Opus 4.6 vs Codex 5.3 We’re building Gandalf and Saruman and naming them like Excel updates.
112
The world is changing. Not tomorrow. Now. The comfort zone? Deadly. AI early adoption? Non-negotiable. Waiting for the next Copilot update? Too late. Move or get moved. What AI tool changed your workflow this year?
1
1
69
Every prompt hitting your most expensive model is like taking a Ferrari to buy milk. LLM routing: half the cost. Same quality.
2
1
64
AI runs on data. I run on panic and coffee. ☕⚡ There’s no model in the world that can replicate what I do in the last 30 minutes before a deadline. Procrastination isn’t a flaw. It’s a compression algorithm for productivity. While AI needs millions of tokens to perform, I just need one thing: A due date in 30 minutes. 😤
1
1
4
79
AI can teach you anything. How things work. Why things break. What to do next. But nobody asks "explain this to me." Everyone asks "do this for me." Then one day: "Why is AI replacing us?!" Bro, you handed it the keys.
78