I would like to ask a question about the difference between Agent Builder and Agentic AI.
If I have a scenario where the agent needs full access to all database records—such as in a car workshop system, where a question could be: “What is the oldest car?”—then the agent must have complete visibility over the database.
My questions are:
Is it better to use Agent Builder and connect it directly to the application’s database as a data source? Or should I use Agentic AI and ground the data inside the flow?
And in the case of using Agent Builder and creating an agent:
Does the agent actually scan all the data?
Or does it generate a query and fetch the needed records directly from the database? This point is important because it impacts performance and speed.
Your experience and insights on these points would be greatly appreciated.
From what I understand, Agent Builder mainly allows the agent to retrieve data from the database for the user.
Meanwhile, OutSystems’ agentic AI can update data directly in the database.
If your use case requires making direct updates to the database, then using agentic AI would be the better option.
It gives you more control because the AI actions are executed through a workflow, so the AI is not given unrestricted access. This way, you can monitor and control the AI’s operations more safely and effectively.