#llmplanning

There are two major paradigms in LLM-based agent design:

Prompt chaining: Static sequences of prompts

Autonomous planning: Agents generate subgoals, tools, and plans dynamically

While prompt chaining is easy to debug, it lacks flexibility. Planning-based agents adapt on the fly, deciding what to do next based on context and feedback.

For frameworks that support both paradigms, visit the AI agents resource page.

There are two major paradigms in LLM-based agent design:

Prompt chaining: Static sequences of prompts

Autonomous planning: Agents generate subgoals, tools, and plans dynamically

While prompt chaining is easy to debug, it lacks flexibility. Planning-based agents adapt on the fly, deciding what to do next based on context and feedback.

For frameworks that support both paradigms, visit the AI agents resource page.