Crack the Agent Mind

12345678910111213141516171819
Across
  1. 2. Retrieval-based technique that augments LLM responses with external knowledge
  2. 3. Technique enabling LLMs to fetch and ground responses in real data
  3. 6. Structured instruction given to an LLM
  4. 8. USE Ability of LLMs to interact with APIs and external systems
  5. 9. Event or signal that initiates an AI workflow
  6. 12. Numerical representation of text for AI models
  7. 13. Process of measuring and validating LLM output quality
  8. 15. Step where the system analyzes input and decides next action
  9. 18. End-to-end flow connecting inputs, models, tools, and outputs
  10. 19. DB Database optimized for similarity search using embeddings
Down
  1. 1. Framework for building stateful, multi-step AI agents
  2. 4. System capable of autonomous decision-making and task execution
  3. 5. Short-term memory used during LLM interactions
  4. 7. REASON ACT Loop followed by intelligent agents
  5. 10. Reusable component in AI systems (retriever, memory, tool)
  6. 11. Stage where the model generates output from input
  7. 14. Framework used to chain prompts, tools, and models
  8. 16. Autonomous system that acts without continuous human input
  9. 17. TUNING Improving model performance using feedback or training data