PinnedInTowards AIbyLouis-François Bouchard·Feb 26Most AI Agents Break in ProductionWe built a course for engineers who are tired of demos and want systems that actually survive deploymentA response icon5A response icon5
PinnedInTowards AIbyLouis-François Bouchard·May 13, 2025Become an LLM dev in 50 hours — learn, code, ship, and certifyMaster the full LLM stackA response icon2A response icon2
PinnedLouis-François Bouchard·Jul 13, 2021About meHi, I am Louis (loo·ee, French pronunciation), from Montreal, Canada, also known as What’s AI on YouTube or medium. I am focusing on…A response icon1A response icon1
InTowards AIbyLouis-François Bouchard·Jun 26How to Build a Memory Your AI Agents Can Actually ReuseThe useful part is not giving agents more context. It is making your research, notes, and sources available again in the next session.A response icon3A response icon3
InTowards AIbyLouis-François Bouchard·Jun 23Why Microsoft Trained MAI-Thinking-1 Without Synthetic DataThe useful lesson is not only that Microsoft released a reasoning model. It is what the team says it filtered out, avoided, and refused to…A response icon1A response icon1
InTowards AIbyLouis-François Bouchard·Jun 15Don’t Put This API in Your AgentCheap proxies can rewrite the workA response icon5A response icon5
InTowards AIbyLouis-François Bouchard·Jun 10Loop Engineering: The New Way to Use Claude Code & CodexWhat happened in 2026 exactly?!A response icon14A response icon14
InTowards AIbyLouis-François Bouchard·Jun 7How Image Generation Actually WorksThe real process behind Midjourney, Gemini, FLUX, and ChatGPTA response icon5A response icon5
InTowards AIbyLouis-François Bouchard·Jun 5Learn AI Engineering in 2026Start AI Engineering in 2026 — Build real AI systems, mostly for free!A response icon8A response icon8
InTowards AIbyLouis-François Bouchard·Jun 3Vibe Coding Only Works If You Already Know How To CodeThe dangerous version of vibe codingA response icon8A response icon8