AI Explored Rethinking Prompting: Getting AI to Work for You
205 snips
Feb 10, 2026 
Jordan Wilson, AI strategist and founder of Everyday AI who teaches context engineering and prompt workflows. He explains why treating models as collaborators beats using them as search. Learn about priming, the Prime-Prompt-Polish framework, using examples to correct outputs, and practical tips for building reliable, maintainable AI assistants.
AI Snips
Chapters
Transcript
Episode notes
LLMs Are More Than Search Replacements
- Jordan argues that treating LLMs like smarter search engines wastes their potential and yields mediocre results.
- He compares it to owning a Ferrari and using it as an umbrella to highlight misuse.
Accept Models Can Outperform You
- Embrace that frontier models can outperform humans when used correctly; this shift is uncomfortable but necessary.
- Treat models as collaborators in an augmented intelligence loop to amplify both model and human performance.
Pick The Right Model For The Task
- Test multiple models and choose the right one for each problem instead of relying on a single system for everything.
- Prefer 'thinking' models for hard queries and avoid defaulting to faster, lower-quality modes.