OpenAI
AIIntegrate OpenAI's full platform: the Responses and Chat Completions APIs with structured output and web search, embeddings, DALL-E and gpt-image-1, Whisper and TTS, files, batches, vector stores, and video.
What This Integration Enables
The OpenAI connector spans the whole platform. The Responses and Chat Completions APIs handle generation, structured output, reasoning models, and built-in web search. Embeddings and vector stores power retrieval. Image, audio, and video cover multimodal generation and transcription. Files, batches, and moderation support the operational pieces around a production workflow. Having the full surface behind one connector means an agent can chain classification, generation, retrieval, and safety checks without stitching vendors together. The value is not any single call but the sequence, and the human review point built into it. An orchestration layer owns that sequencing and gating, and FlowRunner is built for that layer.
Without FlowRunner
With FlowRunner
Use Case Scenarios
Drafted Support Replies
A support email lands in the shared inbox. The agent classifies the intent, drafts a grounded reply with the Responses API, and runs the draft through moderation before it goes anywhere. It stages the reply against the ticket for the support agent, who edits and approves before it sends. Support starts from a solid draft instead of a blank reply box, and nothing goes out unreviewed.
Retrieval-Grounded Q&A
Employees need answers from internal content. The agent embeds the corpus into a vector store, retrieves the relevant passages for each question, and generates a grounded answer with citations. Confident answers return directly; questions touching sensitive policy route to the owning team, so self-serve never guesses on something that matters.
Multimodal Content Production
A campaign needs copy, images, and a voiceover. The agent drafts copy with a reasoning model, generates image options with gpt-image-1, and produces a voiceover with TTS. Everything goes to a person for selection and approval before publishing, so production speeds up while a human owns what ships.
Human-in-Loop Highlight
A generated reply, answer, or asset is a draft the flow produces, not a final decision. When OpenAI output will reach a customer, commit the company, or touch sensitive material, FlowRunner routes it through a [human-in-the-loop](/concepts/human-in-the-loop/) step: the agent pauses, presents the draft with its context and any moderation result, and sends it to the responsible person via Slack. They edit and approve. The full platform produces the work; a person signs off before it leaves the flow.
Agent Capabilities
39 actionsModeration
1- Moderate Content Analyzes text and image inputs for harmful content across multiple safety categories (harassment, hate speech, violence, sexual content, self-harm, and more) using OpenAI's Moderation API.
Audio
3- Text to Speech Converts text into natural-sounding speech audio using OpenAI's text-to-speech models. Uploads the generated audio file and returns its URL.
- Speech to Text Transcribes an audio file into text using OpenAI's speech recognition models. Downloads the audio from the provided URL and returns the transcribed text.
- Translate Audio Translates spoken audio in any supported language into English text using OpenAI's Whisper model.
Web Search
1- Web Search Generates a grounded answer to a prompt by letting the model search the web for current information.
Responses
5- Create Response Generates a model response via OpenAI's Responses API, the primary interface for text generation.
- Get Response Retrieves a previously created model response by its ID, including its status, output items, and token usage.
- Cancel Response Cancels an in-progress model response that was created with Background enabled. Only background responses can be cancelled.
- Delete Response Permanently deletes a stored model response by its ID from OpenAI's servers.
- List Response Input Items Lists the input items (messages, tool calls, etc. ) that were used to generate a stored response, with cursor-based pagination.
Chat
1- Create Chat Completion Generates a model response via OpenAI's Chat Completions API from a full messages array (system/user/assistant/tool roles, including vision image content parts).
Embeddings
1- Create Embeddings Generates vector embeddings for one or more text inputs using OpenAI's embedding models. Supports the 'dimensions' parameter of the text-embedding-3 family to produce shortened vectors.
Images
2- Generate Image Generates images from a text prompt using OpenAI's image models ('gpt-image-1' or the DALL·E family).
- Edit Image Edits or extends existing images from a text prompt using OpenAI's image models. With 'gpt-image-1', multiple input images can be combined into a new composition; an optional mask marks the transparent areas to be repainted (inpainting).
Files
5- Upload File Uploads a file to the OpenAI account by downloading it from the provided URL. Uploaded files can be used across the API: batch inputs (purpose 'batch'), file search / vector stores (purpose 'assistants'), vision inputs (purpose 'vision'), fine-tuning datasets (purpose 'fine-tune'), and general Responses API inputs (purpose 'user_data').
- List Files Lists the files uploaded to the OpenAI account, optionally filtered by purpose, with cursor-based pagination and sorting by creation date.
- Get File Retrieves the metadata of an uploaded file by its ID, including filename, size in bytes, purpose, and creation date.
- Delete File Permanently deletes an uploaded file from the OpenAI account by its ID.
- Download File Content Downloads the content of a file stored in the OpenAI account (e.g. a batch output or error file), uploads it to FlowRunner file storage, and returns its URL.
Batches
4- Create Batch Creates a batch job that processes a . jsonl file of API requests asynchronously within 24 hours at 50% of the standard cost.
- Get Batch Retrieves the current state of a batch job by its ID, including status (validating, in_progress, completed, failed, expired, cancelled), request counts, and the output_file_id and error_file_id for downloading results once finished.
- List Batches Lists the batch jobs in the OpenAI account with their statuses and request counts, using cursor-based pagination.
- Cancel Batch Cancels an in-progress batch job by its ID. The batch moves to 'cancelling' for up to 10 minutes and then 'cancelled'; partial results of already-completed requests remain available in the output file.
Vector Stores
8- Create Vector Store Creates a vector store for semantic file search (RAG). Optionally attaches already-uploaded files (purpose 'Assistants'), which are automatically chunked and embedded.
- List Vector Stores Lists the vector stores in the OpenAI account with their statuses, file counts and storage usage, using cursor-based pagination.
- Get Vector Store Retrieves a vector store by its ID, including its status, file counts, storage usage, and expiration policy.
- Delete Vector Store Permanently deletes a vector store by its ID. The underlying uploaded files are not deleted and remain in the account.
- Add File to Vector Store Attaches an already-uploaded file to a vector store. The file is automatically chunked and embedded; poll 'List Vector Store Files' or 'Get Vector Store' until processing completes.
- List Vector Store Files Lists the files attached to a vector store with their processing statuses, optionally filtered by status, using cursor-based pagination.
- Remove File from Vector Store Detaches a file from a vector store by its ID. The underlying uploaded file is not deleted and remains in the account.
- Search Vector Store Performs a semantic similarity search over a vector store and returns the most relevant file chunks with their scores, a direct RAG retrieval step without invoking a model.
Videos
6- Create Video Starts an asynchronous video generation job from a text prompt using OpenAI's Sora models.
- Remix Video Creates a new video generation job based on a completed video, applying the changes described in the prompt while preserving the original's structure.
- Get Video Retrieves a video generation job by its ID, including its status (queued, in_progress, completed, failed) and progress percentage.
- List Videos Lists the video generation jobs in the OpenAI account with their statuses, using cursor-based pagination.
- Delete Video Permanently deletes a video generation job and its assets from OpenAI's servers by its ID.
- Download Video Content Downloads the rendered content of a completed Sora video job (the MP4 video, a thumbnail image, or a spritesheet), uploads it to FlowRunner file storage, and returns its URL.
Models
2- List Models Lists all models currently available to the account, including their owners and creation dates.
- Get Model Retrieves the basic details of a specific model by its ID, confirming that it is available to the account.
Start building with OpenAI
$100 in credits. No card required. Connect in minutes.