ChatGPT

Connect ChatGPT to use OpenAI models for AI chat, Message Templates, and flow blocks. ChatGPT can generate or analyze text, while a separate email or text-message integration sends completed messages.

  • You want to use OpenAI models in AI chat, Message Templates, or a flow.
  • You want structured text analysis features backed by OpenAI models.
  • You want to choose a model for an AI chat or individual flow block, or use the default model.
  • Create or copy an OpenAI API key from the OpenAI platform.
  • Understand that model availability can depend on your OpenAI account, permissions, and billing status.
  • Confirm your role can create integrations. Without that permission, New Integration is not shown.
  • Confirm Use AI on this account is on. Only members with settings edit permission can change it.
Choose ChatGPT in the AI integration wizard.
  1. In the Dashboard sidebar, select Integrations, then AI.
  2. Select New Integration.
  3. Choose ChatGPT.
Review the ChatGPT connection fields before verifying your OpenAI API key.
  1. Review the Name, which defaults to “ChatGPT,” and enter the OpenAI API Key.
  2. Select Verify API Key. Drip Drop saves the integration, asks OpenAI to validate the key, and loads the available models.
  3. After validation succeeds, optionally choose a Default Model, then select Save & Continue.

When an AI flow block does not select a model, ChatGPT uses an available non-mini GPT-4o model when one exists; otherwise, it uses the first supported model returned by OpenAI. The form recommends GPT-4o for a balance of quality and speed when it is available. Review OpenAI model documentation if you expect a specific model family.

Note
AI chat requires a model that supports tool calling. If ChatGPT rejects a request, choose a different model from the Model menu in AI chat.
  1. On the Verify step, review the saved-credentials status. To run another live OpenAI validation, select Check credentials.
  2. Before the second check, Credentials stored confirms only that credentials are saved. After Check credentials succeeds, it reports “These credentials work.”
  3. Select Done.
  • In AI chat, choose ChatGPT as the provider, then ask about your account or the documentation. If model selection is available, use Model to choose one for the conversation. Suggested changes wait for your approval.
  • In Message Templates, open AI Assistant, choose ChatGPT, and ask for a draft, rewrite, or tone change. Apply or dismiss each proposed change from its preview.
  • In the Flow Editor, select this integration under AI Integration and choose Write & Send Response. Select a separate Sending Integration for email or text-message delivery.
  • In a flow AI block, choose Analyze Response to read text and save structured results without sending a message.
  • When a flow template asks for an AI integration, select the ChatGPT integration for that slot.
API key and authentication problems
  • If the provider does not validate, confirm the OpenAI API key is correct and still active.
  • If the key looks right but access still fails, check whether the OpenAI account has billing, quota, or permission restrictions.
Model availability problems
  • If no models appear, confirm the provider validated successfully and the OpenAI account can list models.
  • If the expected model is missing, remember that Drip Drop only shows supported chat-capable model families in this picker.
  • If a model selected in a flow stops working, choose a different option under Model (optional) and retry the flow.
Rate limits and provider errors
  • If AI actions fail intermittently, OpenAI rate limits, quota issues, or temporary service problems may be the cause.
  • If the provider returns a bad-request or permission error, re-check the selected model and API access rather than treating it like a generic network problem.
  • See Integrations for the shared provider setup flow and provider management basics.
  • Compare Claude if you want a similar AI-provider workflow using Anthropic models instead of OpenAI models.