LLM agents

Large language models allow programs to reason beyond fixed rules. By combining prompts, state, and tool use, we can build agents that plan, adapt, and respond flexibly. These techniques introduce probabilistic decision-making into structured systems.

• LLM reasoning — Using language models for decision-making.
• OpenAI Responses API — Making structured model calls from Python.
• State and memory — Supplying and updating contextual information.
• Tool selection — Letting models choose and invoke functions.
• LLM workflows — Planning and executing multi-step tasks.
• Single-agent loop — Integrating LLM reasoning into the agent cycle.