OpenAI connector icon

OpenAI Connector

OpenAI is an artificial intelligence research and deployment company focused on creating advanced AI systems intended to "benefit humanity".

01

Fast in-app AI features

Expose agent-ready capabilities such as summarization, ticket triage, or content generation without building a…

02

Governed agent access (MCP)

Turn workflows and approved connector methods into MCP tools or MCP-triggered workflows so AI…

03

Webhook and background job handling

Use OpenAI webhook events (for example response.completed and batch.completed) to trigger workflows or notify…

04

Embedded experience

Deliver customer-facing workflows natively inside your SaaS product.

CONNECTOR OVERVIEW

About this connector

The Cyclr OpenAI Connector lets your product make authenticated calls to the OpenAI API OpenAI documentation using an API key. Common API surface areas include the Responses API, chat and conversation endpoints, embeddings, moderation, files, audio and realtime features. OpenAI also supports webhook events for background responses and batch jobs which you can subscribe to and route into Cyclr workflows. OpenAI documentation

PRODUCT USE CASES

Practical use cases for multi-tenant SaaS products

Cyclr helps product teams turn OpenAI API capabilities into governed, tenant-scoped features inside your app. Use the connector to:

Agent-safe in-app assistants

Expose curated capabilities (summarize tickets, draft replies, pull customer context) as MCP tools rather than raw API calls so LLM agents run governed workflows that respect tenant credentials, redaction and audit trails. Cyclr MCP-triggered workflows let you package multi-step logic and control exactly what data is returned to the model. OpenAI documentation

RAG-enabled knowledge widgets

Create embeddings from customer documents, store vectors in a vector store and build retrieval-augmented generation (RAG) flows. Use Cyclr to orchestrate the embedding, vector store upsert and response generation steps, keeping each customer’s data isolated. OpenAI documentation

Automated moderation pipelines

Run OpenAI moderation on user generated text or images before publishing. Use Cyclr to route flagged content into review queues, apply policy rules, notify admins or block publishing — all within tenant boundaries. OpenAI documentation},{

ORCHESTRATION

How OpenAI works with Cyclr

Cyclr sits between your SaaS product and OpenAI to handle authentication, govern access and orchestrate every API call per customer, at scale.

  1. 1

    Your SaaS product

    Your app requests an action for a specific customer.

  2. 2

    Cyclr orchestration

    Cyclr authenticates, applies policies and routes the request.

  3. 3

    OpenAI

    OpenAI processes the request and returns a response to your app.

Customer-specific authentication · Governed data movement · Centralized visibility and audit

CONNECTOR COVERAGE

Available connector capabilities

Explore OpenAI methods you can orchestrate through Cyclr.

Model Responses2

Delete Model Response

Action

Delete a model response.

Create Model Response

Action

Create a model response.

Embeddings1

Create Embeddings

Action

Create embedding vectors representing input text.

TECHNICAL DETAILS

OpenAI integration and API details

Review the sourced authentication and API behaviour used when configuring customer-facing OpenAI integrations in Cyclr.

API and authentication

API host
https://api.openai.com/v1
Authentication
API key (HTTP Bearer) with optional OpenAI-Organization header
Credentials

Typical Cyclr connector fields: API Key (required) and OpenAI Organization (optional). Cyclr’s guide shows the same fields when installing the connector. Usage is billed…

Method categories

Common API categories available via OpenAI’s public API include Responses and chat/conversation endpoints, Embeddings, Images, Audio and speech, Files and uploads, Fine-tuning and batches…

View vendor API documentation →

Pagination and rate limits

Pagination

List endpoints typically use cursor-style pagination (for example after + limit) and may return pagination metadata (firstid, lastid, hasmore). Use those parameters when listing resources such as files, projects or users. OpenAI documentation

Rate limits

OpenAI enforces rate limits that vary by model, endpoint and your subscription. The API surfaces debugging and rate information in response headers; do not rely on a single static number. Design retries, backoff and graceful…

IMPLEMENTATION

Authentication at a glance

OpenAI authentication uses API keys provided via the HTTP Authorization header (Authorization: Bearer ). If you need to scope requests to a specific organization or project, include the OpenAI-Organization and/or OpenAI-Project headers. Keep keys secret and store them in Cyclr’s connector credential store for tenant…

Authentication method
API key (HTTP Bearer) with optional OpenAI-Organization header
Credential required

Typical Cyclr connector fields: API Key (required) and OpenAI Organization (optional). Cyclr’s guide shows the same fields when installing the connector. Usage is billed…

Before you begin:

OpenAI API access is a paid service; usage is billed to the organization tied to the API key. Make sure customers understand where usage…

View setup guide

COMMON QUESTIONS

Frequently asked questions

The connector requires an OpenAI API key (Authorization: Bearer ). An OpenAI Organization ID is optional and used to scope usage to a specific organization or project. Cyclr’s setup guide shows these fields when installing the connector. OpenAI documentation

Store the API key in Cyclr’s connector credential store for account-level isolation. Cyclr uses that key when making API requests to api.openai.com and can include the OpenAI-Organization header when provided. Do not embed API keys in client-side code. OpenAI documentation

Yes. Use Cyclr to expose connector methods or entire templates as MCP tools or MCP-triggered workflows. That lets agents call governed workflows instead of raw APIs so you can enforce input validation, redaction, authentication and audit logging. OpenAI documentation

Yes. OpenAI publishes webhook events such as response.completed, response.failed and batch.completed. You can configure a webhook endpoint and have Cyclr ingest those events to continue or finalize workflows. SDKs include helper methods to verify webhook signatures. OpenAI documentation},{