Hallo @smymisr

From my experience, I hope this information is helpful to you 

What’s your experience building or integrating AI agents (e.g., chatbots, copilots, decision engines) within OutSystems?

It was an enjoyable and smooth experience. The performance and output quality depend on choosing the right AI model for your needs. For example, if you need text processing or image analysis, you must select the model that fits the requirement.

How does ODC handle real-time inference, prompt orchestration, or external AI service integration (e.g., Azure OpenAI, Google Vertex AI, LangChain)?

You should use settings for ease of use and organize them into reusable groups. For example:

• Create a property for email formatting that can be used everywhere.
• Another property for color usage in reports, and so on.

Regarding data, send only the necessary data to reduce token consumption and cost.

Are there best practices for managing stateful conversations, context memory, or multi-turn logic in ODC?

You need to know what fits the task. For example:

• In chat scenarios, use the Save Memory Data option to maintain conversation consistency.
• In other cases, like generating a report, do not save memory if it’s not needed.

How does ODC compare to other low/no-code platforms (e.g., Mendix, Power Apps, Appian, Retool) in terms of AI integration cost, scalability, and developer control?

Cost depends on the type of organization:

• Large banks or insurance companies can afford licensing costs and benefit from advanced features, making OutSystems or Mendix suitable.
• Small companies cannot bear high costs and will rely on AI tools with cheaper licenses.

Any insights on licensing implications when embedding external AI services or using custom connectors?

I’m not aware of this, my friend.