Édouard Kely Presley Germain
-
V1.2.1: I Read My Own Post and Realised I Was Wrong
What happens when you re-read your own code weeks after writing it, with no open bug and no pressure. Sometimes that is all it takes. Read more
-
v1.7: Single Agent vs Multi-Agent: What I Learned Building Both
Single agent or multi-agent? Emergent or explicit routing? Here’s what I learned with an AI system that started too rigid and had to evolve. Read more
-
v1.6: What Happens When You Stop Building What You Know
What happens when you stop building what you know: a video game, a ugly website, and an AI dashboard Read more
-
v1.5: I Built an NFL Draft Model With the Same Stack I Use at Work
I’ve spent the past few weeks building data pipelines and machine learning models. Not for work—for the Los Angeles Rams. On weekends. For free. This probably needs some context. I build AI agents, wrangle data, and occasionally argue with LLMs about what year it is—because I’m paid to, and because I genuinely enjoy it. But… Read more
-
v1.4: The Hallucination That Almost Made It to Production
Recently, I have been working on some new things that I always wanted to catch up on. In this AI space, it always feels like you’re way behind on concepts! But behind the constant flow of inventions (and sometimes disguised reinventions), there are very important core concepts that are interesting to know in order to… Read more
-
v1.3: Three Days Arguing With an LLM About What Year It Is
I once built a data analysis agent. Simple stuff – query the database, answer questions about transactions and campaigns. User asks something, agent writes SQL, runs it, gives an answer. Worked great until I asked: “Show me our Q1 2025 performance.” The agent pulled the data perfectly. Then told me: “This is a forecast for… Read more
-
v1.2: What Building Access Control for AI Agents Taught Me
I was building an access control system for a multi-tenant application. Pretty standard stuff – internal employees get instant access, external partners need invitations. We already had Azure AD authentication working, just needed to automate adding people to the security group instead of having the IT team do it manually through the portal. I figured… Read more
-
v1.1: Why does AI code „always“ fail in production?
AI accounts for over half the code produced in some organizations, yet 48% of AI-generated code snippets contain vulnerabilities compared to lower rates for human-written code. The industry narrative is clear: AI-generated code is fundamentally less reliable in production. But what if we’re blaming the wrong thing? AI-generated code doesn’t come with a self-destruct mechanism.… Read more
-
v1.0: Hello World!
Hello everyone! My name is Edouard. The first post is always the hardest, but I thought I’d start writing about anything and the rest would come. 276 was the telephone prefix for my hometown (on an island far, far away…) and starting a blog, I thought I would use it in the name, to keep… Read more
276AI 