Enabling generative AI

Generative AI in Business Automation Workflow uses large language models (LLMs) to automate content‑generation tasks and improve workflow productivity by streamlining content‑based processes. You can enable generative AI in Business Automation Workflow either by using the IBM Model Gateway to access multiple LLM providers or by connecting to watsonx.ai directly.

Model gateway connection
Connecting through the model gateway gives you access to multiple LLM providers, including IBM watsonx.ai, OpenAI, Google Gemini, and AWS Bedrock. This approach provides vendor-agnostic access with centralized governance and security. Due to its flexibility and the advantages it offers, the gateway method is the preferred approach.
Direct watsonx.ai connection
Connecting directly to IBM watsonx.ai as your LLM provider is suitable when you want to use watsonx.ai exclusively and do not require access to other LLM providers. See Enabling generative AI with IBM watsonx.ai.

Choosing the appropriate approach

To decide which approach to use, consider the following factors.

  • Use the model gateway connection when:
    • You need access to multiple LLM providers through a single interface, with ability to choose the best LLM for each specific use case.
    • You want to maintain vendor flexibility, avoid lock‑in, and plan to easily adopt new providers.
    • You need centralized authentication, policy enforcement, and audit logging.
    • You want to optimize costs by using different providers for different workloads.
  • Use direct watsonx.ai connection when:
    • Your organization has standardized exclusively on watsonx.ai for all AI workloads.
    • You're working in a highly constrained environment where the AI gateway cannot be deployed.
    • You need a simpler configuration with fewer components, for example, when you're implementing a proof-of-concept project with limited scope.

After you choose the approach that best fits your requirements, follow the corresponding setup instructions.

For generative AI usage information, see Adding a generative AI task to a service flow and Building a generative AI task in a service flow.