Tool Use - AI Conversations Fundamentals of AI Tooling (ft Hai Nghiem)
13 snips
Aug 26, 2025 Hai Nghiem, co-founder of AI Ventures Canada and educator in AI fundamentals, dives deep into the world of AI tooling. He explains the concept of context engineering as a superior alternative to prompt engineering. Listeners will explore the distinctions between AI agents and chatbots, and learn about the Model Context Protocol— likened to 'USB-C for agents.' Nghiem also shares practical tips for beginners using coding tools, highlighting the power of deep agents in enhancing productivity and reasoning. Perfect for anyone looking to navigate the AI landscape!
AI Snips
Chapters
Transcript
Episode notes
Context Engineering Is The Core
- Context engineering means controlling what goes into the model's context window, not changing the model itself.
- Prioritize relevant, preprocessed information over dumping raw logs to improve LLM output quality.
Agent Vs Workflow Distinction
- Agents are open-ended systems that form plans on the fly, like a person solving vague tasks.
- Workflows are predictable step sequences and suit tasks that require consistent, high-quality outputs.
Build Agents From Expert Workflows
- Talk to the human who does the job and model their decision process into the agent's context and prompts.
- Iterate by running the agent on real tasks and curate a golden dataset of human judgments to improve outputs.