AI Glossary
Essential concepts that transform how you communicate with AI. Each term gives you more control and clearer results.
Deconstruct
Break down complex logic or code into its fundamental components. Instead of asking AI to 'figure it out,' explicitly outline each step and requirement.
Examples:
"Deconstruct this function line by line to identify the failure point."
Lifecycle
Design the complete flow from initialization to cleanup. Specify startup, operation, and shutdown phases explicitly.
Examples:
"Design the complete lifecycle for database connections: startup, connection pooling, query execution, and graceful shutdown."
Hallucination Audit
Verify that each function works as claimed. Force the AI to check its own work and identify logical inconsistencies.
Examples:
"Conduct a hallucination audit on this implementation: verify each function works as claimed and identify any logical inconsistencies."
Agentic Coding Cycle
A three-phase approach: Plan (lay out the approach), Execute (implement fully), Audit (verify correctness). Transforms AI from guessing machine to systematic problem solver.
Examples:
"Plan: Lay out a plan to refactor this authentication system. Execute: Execute the plan in full per TypeScript best practices. Audit: Conduct a hallucination audit on the refactored code."
Recall
The ability to retrieve and reference previously established context, patterns, or solutions. Use recall to maintain consistency across interactions.
Examples:
"Recall the authentication pattern we established earlier and apply it here."
Refactor vs Scaffold
Refactor improves existing code structure without changing behavior. Scaffold creates new structure from scratch. Know when to use each approach.
Examples:
"Refactor this module to improve maintainability without changing its API."