The platform's AI agents and skills call models through integrations you register under Platform Admin → AI Models / Integrations. Each integration names a model, its provider and the provider's base URL, and holds the API key (your provider token) the platform uses to call it. Register your keys before anyone launches a campaign with the AI assistant: until then, Launch a New Campaign can only be completed by writing the brief yourself. Every call runs from inside your own AWS account.

Before you begin

  • An HQ Admin account with the Admin role.
  • An API key from each provider you will use, created in that provider's own console with access to the model you register:
    • OpenAI (ChatGPT / GPT) — writes the brief, storyboard and copy.
    • Google Gen AI (Gemini) — generates the campaign images.
    • Google Gen AI (Veo 3.1) or LTX (LTX 2.5) — generates video. Register 1 of the 2; you do not need both.

Which integrations you need

StageProviderExample modelNeeded for
Brief, storyboard and copyOpen AIgpt-5.2Every campaign created with the AI assistant
ImagesGoogle Gen AIgemini-3.1-flash-image-previewImage campaigns, and the scene frames of a video campaign
VideoGoogle Gen AI or LTXveo-3.1-generate-preview or ltx-2-5-proVideo campaigns only

Register 1 integration per row. A brief written by the assistant with your own uploaded creative needs the first row only.

Add an integration

  1. In the left navigation, select Platform Admin.
  2. Under Enabled Features, select AI Models / Integrations.
  3. Select + Create Integration.
  4. Enter an Integration Name you will recognise when assigning it to a skill, for example GPT 5.2 or Veo 3.1.
  5. Choose the Provider: Open AI, Google Gen AI or LTX.
  6. Choose the Model, for example LTX 2.5 Pro for LTX. The integrations table lists each model by its identifier, such as gpt-5.2, gemini-3.1-flash-image-preview, veo-3.1-generate-preview or ltx-2-5-pro.
  7. If the provider needs one, enter the Base URL, for example https://api.openai.com/v1 for Open AI or https://generativeai.googleapis.com/v1 for Google Gen AI. LTX needs no base URL.
  8. Paste the provider's key into API Key.
  9. Select Save Changes.

Repeat for each row in Which integrations you need.

Check the skill assignments

  1. Under Enabled Features, select AI Agents / Skills.
  2. On the System Prompts tab, read the AI Model / Integration column.
  3. Where a skill names an integration you have not registered, assign one of yours. AI agents and skills reference lists the stage each skill belongs to.

Set retry behaviour

  1. Back under AI Models / Integrations, in the Integration Settings card, enter Max Retries — how many times a failed model call is retried, for example 3.
  2. Enter the Retry Interval between retries in seconds, for example 2.
  3. Select Save changes.

Verify it worked

Each integration appears as a row in the table with its Name, Model, Provider and Base URL (blank for LTX), and is available to choose in the AI Model / Integration column under AI Agents / Skills. Under Campaigns → Launch a New Campaign, the AI assistant now writes a brief from a typed or spoken description.

Troubleshooting

Campaign brief or creative generation fails

Cause. The model's API key is invalid, or the base URL or model identifier does not match what the provider expects.

Fix. Open the integration from the ⋮ menu in the Actions column, check the Model and Base URL against the provider's documentation, re-enter the API Key, then select Save Changes.

Video generation fails while images work

Cause. No video integration is registered, or the video skill is assigned to an image or text model.

Fix. Register Veo 3.1 (Google Gen AI) or LTX 2.5 (LTX), then assign it to the video skill under AI Agents / Skills.

Next step

Launch a new campaign