Agentic AI, LLM engineering, and the models behind modern automation — multi-agent systems, LLM evaluation and comparisons, RAG, fine-tuning, AI infrastructure, security, and production AI engineering.
From the blog
Learn how to install the OpenAI Agents SDK, configure your API key, create your first intelligent agent, and run it with Runner.run_sync(). A complete hands-on tutorial.
Learn how to handle user interruptions and barge-in events in voice agents with lifecycle management, audio muting, graceful cancellation, and response resumption strategies.
Use HostedMCPTool to run MCP tools server-side on OpenAI's infrastructure with zero callback overhead, including tool_config structure, approval options, and GitMCP for repository access.
Learn how to use OpenAI's hosted tools — WebSearchTool, CodeInterpreterTool, FileSearchTool, and ImageGenerationTool — to give your agents powerful built-in capabilities without writing custom logic.
Combine the handoff pattern with the agents-as-tools pattern to build hybrid orchestration systems where a triage agent delegates via handoffs and specialist agents use sub-agents as tools.
Learn how to implement input guardrails in the OpenAI Agents SDK to validate, filter, and reject unsafe user requests before they reach your agent's main processing loop.
Understand the core architecture of voice AI agents — STT to Agent to TTS pipelines, the VoicePipeline SDK approach vs the Realtime API WebRTC approach, and when to use each for production voice applications.
Integrate Anthropic, Google, Mistral, and other LLM providers into OpenAI's Agents SDK using LiteLLM's unified interface with LitellmModel, provider prefix notation, and cross-provider tracing.
Use MCP prompt resources to dynamically load and parameterize agent instructions from external servers, enabling centralized prompt management with list_prompts and get_prompt.
