Full Workshop: Build Your Own Deep Research Agents - Louis-François Bouchard, Paul Iusztin, Samridhi

Building production-ready AI engineering systems requires a strategic balance between simple workflows and complex agentic architectures. Rather than defaulting to multi-agent systems, developers should prioritize the simplest solution—often a deterministic workflow—to maintain control and reduce context rot. A robust deep research agent leverages tools like Gemini and MCP to scrape web content, analyze YouTube transcripts, and synthesize cited reports, while a separate, constrained writing workflow utilizes an evaluator-optimizer loop to produce high-quality, human-like LinkedIn content. Effective AI engineering hinges on rigorous observability and evaluation, where LLM judges calibrated against domain-expert datasets ensure performance consistency. By modularizing capabilities into tools and skills, engineers can create scalable, maintainable systems that automate technical content production without sacrificing quality or reliability.