What's AI by Louis-François Bouchard
LLM Architecture, AI Agents, and RAG with a focus on engineering trade-offs and practical implementation.
Nutrition Label
Louis-François Bouchard delivers high-signal educational content focused on the engineering realities of building LLM applications. He excels at demystifying hype by contrasting theoretical AI capabilities with practical architectural trade-offs learned from real-world client projects. Viewers can expect clear mental models for complex topics like agents versus workflows, often supplemented by academic context.
Strengths
- +
- +
- +
Notes
- !He frequently promotes his own 'Towards AI' courses and books; check descriptions for these commercial links.
- !Content targets engineers and architects making system design choices rather than casual users looking for tool reviews.
Rating Breakdown
Breakdown across the key dimensions we rate. Methodology →
Recent Videos
How AI Agents Actually Work: ReAct vs Plan-and-Execute
The AI Engineer's Dilemma - Choose the Right AI System
Our New Agentic AI Engineering Course!
42 AI Concepts You Actually Need to Understand LLMs
The 12 Questions That Decide Your AI Architecture
Stop Overengineering: Workflows vs AI Agents Explained
5 Edits That Instantly Make AI Text Sound Human
From Workflows to Multi-Agent Systems: How to Choose
Your Prompts Aren’t the Problem—Your Context Is
Prediction Isn’t Understanding and That Difference Matters
Engineering AI Agents — University of San Diego Guest Talk
Agent Skills vs MCP Which Is Better?
OpenAI’s $200/Month Research Agent: Is It Worth It?
Why this rating
Evidence receipts showing why each dimension is rated the way it is.
“We have a new course called Master AI for Work available for you right now on our site.”[28:48] →
Clearly discloses the self-promotional nature of the content (his own course) within the educational context.
“We pivoted a few times. After testing, users almost always used both agents once, without alternating between them... Iterating with the writer agent itself is enough.”[49:22] →
The speaker reveals a specific pivot in architecture based on observing actual user behavior during a proof-of-concept, demonstrating genuine product development experience.
“Token usage: Up to 15x more than single agent... Latency: Coordination takes time... Reliability: More moving parts = more failure points.”[29:33] →
Instead of hyping agents, he explicitly lists the engineering downsides and costs encountered when moving from single to multi-agent systems.
“The video content is exclusively a trailer/pitch for the course named in the title, delivering exactly what was promised.”[0:00] →