OpenAI Agent

Use OpenAI models for chat, structured output, vision, and safety checks.

What it does

• Sends prompts to OpenAI models for chat-style responses, structured data, images, or vision analysis.
• Provides moderation to screen content and embeddings for search or matching.

Common uses

• Draft or summarize text from orders, tickets, or files.
• Extract structured fields from freeform text using the structured output action.
• Create embeddings for semantic search or deduplication.
• Generate or edit images and answer questions about images.

Key actions from the spec

• Chat and text generation: chat completions and freeform text completions with temperature and stop controls.
• Structured output: enforce a JSON schema for reliable field extraction.
• Vision: describe or answer questions about images, and generate or modify images from prompts.
• Embeddings: create vector embeddings from text.
• Moderation: check text against safety policies before sending it onward.

Setup notes

• Requires an OpenAI connection or API key with access to your chosen models.
• Pick models (for example, gpt-4o) and adjust tokens or temperature based on how creative versus precise you need the output.
• Provide images as base64 or a reachable URL when using vision actions.