French agency delivering innovative projects with Lean Startup & full-stack JS.
Editor of React-admin and GreenFrame.
We’re hiring French devs in full remote.
Getting code cleanup right is super important for performance and refactoring. 🧹
Knip.dev is a handy tool for #JavaScript and #TypeScript projects that detects unused files, dependencies, and exports so you can remove them before deployment.
A recent study by Mistral revealed that AI can be a big carbon emitter.
The good news? We can reduce its environmental impact by making conscious decisions.🌱
Here are 3 best practices:
1️⃣ Opt for smaller, task-specific models
2️⃣ Write clear prompts & request short responses
3️⃣ Choose data centers wisely
📚If you want to learn more about AI’s hidden costs, here are some useful resources:
🔗 https://mistral.ai/news/our-contribution-to-a-global-environmental-standard-for-ai
🔗 https://marmelab.com/blog/2025/03/19/ai-carbon-footprint.html
Claude Code is by far the most impressive AI coding assistant I’ve used so far!
Why? Because it isn't limited to bootstrapping small apps like v0. Claude Code can also work with large & complex code bases.
It turns coders into architects & empowers business experts to build applications autonomously.
I shared my experience in my latest blog post👇
https://marmelab.com/blog/2025/07/29/ai-coding-assistants-are-the-new-compilers.html
#AI #ClaudeCode #v0 #GithubCopilot #LLM #SoftwareEngineering
Are you experiencing DNS slowdowns in your @nodejs app?
Caching could be the solution to your problems! ⚡
How? We break it down in our latest blog post. Check it out now:👇
https://marmelab.com/blog/2025/07/28/dns-in-nodejs.html
#NodeJS #WebPerformance #Backend #DNS #Caching #JavaScript #DevTips #WebDev #TechBlog
From zero to 💯!
Shadcn Admin Kit just crossed 100 GitHub stars. ⭐
Thanks for the love — we're just getting started!
Mistral AI just took a bold step towards transparency by publishing the first lifecycle analysis (LCA) of an AI model.
📊 The results? Training Mistral Large 2 (128B parameter) emitted 20,000t CO₂e.
It confirms what many feared:
👉 AI is a massive carbon emitter.
At Marmelab, this issue has been on our radar for a while. That’s why we conducted our own study earlier this year.
🔗 Read it here: https://marmelab.com/blog/2025/03/19/ai-carbon-footprint.html
Mistral AI emitted 20,000 metric tons of CO2 equivalent to train Mistral Large 2, a 128B parameters model.
That's a first: No third-party study on the carbon footprint of a state-of-the-art LLM has ever been published before.
La start-up française d’intelligence artificielle publie avec l’Ademe et Carbone 4 une étude sur l’impact, notamment sur les émissions de CO₂, de l’entraînement et l’usage de ses modèles. Et appelle ses concurrents à faire de même.
AI makes experienced open-source devs… slower?! 🐢
We might feel more productive with AI—but the data says otherwise:
A new METR study found that when devs use AI tools, they take 19% longer than without.
Maybe AI’s real superpower is making us believe it’s super powerful.
🔗 https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/
Our newest open-source project Shadcn Admin Kit now has a website!
Check it out and let us know what you think 👇
François Zaninotto