When creating or editing an Agent, three tabs are shown. The second tab, titled AI & Tools, lets you configure the reasoning behavior of your Agent and specify which additional Agents and Tools it can interact with.
This tab is divided into two main sections:
- AI Configuration
- Agents & Tools to use
This section defines how the Agent processes and generates responses.
Select the LLM your Agent will use. Models are grouped by provider (e.g., OpenAI, Gemini, Anthropic) and offer multiple versions depending on their capabilities—some prioritize speed, while others handle complex requests more effectively.
After choosing a model, this setting is populated automatically. It controls how predictable or creative the Agent’s responses should be. A lower value leads to more consistent outputs; higher values introduce more variation.
Also filled in automatically after selecting a model. This setting determines the maximum length of responses, with higher values allowing more detailed answers.
This setting defines how the AI processes information to generate responses. Depending on the selected strategy, the Agent can produce more logical, creative, or robust answers. The following options are available:
-
Enables complex reasoning by guiding the LLM through intermediate steps. Ideal for problems that require sequential logic or explanation-based answers. More information: Chain of Thought details.
-
Adjusts the prompt in real time based on user input or intermediate results, allowing the Agent to adapt its reasoning dynamically as the conversation progresses. More information: Dynamic Prompting details.
-
Focuses on improving the user’s initial query. The LLM reformulates unclear or ambiguous questions before responding, which leads to more accurate and relevant answers. More information: Question Refinement details.
-
Allows the LLM to review its intermediate reasoning, detect possible mistakes, and revise its output before finalizing the response. This strategy is useful for improving accuracy but should not be used when the expected output is strictly a structured format (e.g., a JSON or a list). More information: Reflection details.
-
Executes multiple reasoning paths for the same query and selects the most consistent result. This approach increases the reliability of the answer in complex or open-ended tasks. Self Consistency details.
-
Explores multiple reasoning paths in parallel by maintaining a tree structure of possible thoughts (language sequences). It selects the best path through iterative evaluation of intermediate ideas. More information: Tree of Thought details.
-
Uses the model without prior examples or guidance. The Agent relies entirely on its pre-trained knowledge to respond directly to the input. More information: Zero-shot details.
-
Combines zero-shot prompting with explicit instruction to "think step by step", triggering more thoughtful, structured reasoning even in tasks with no examples. More information: Zero-shot CoT details.
This section allows you to configure which Agents and Tools will be available for the Agent you are defining. In this way, the Agent can execute specific actions, extend its capabilities through external functions or other Agents already created.
To add a new Tool or Agent, click on the + Add button. This opens a window where you can browse all available Tools and Agents.
In the left pane, you can see a list that includes both Tools and Agents.
Note that you can choose to view:
- Public Tools only
- Private Tools only
- Defined Agents (as they can be used as Tools within the Agent you are creating)
- All of the above
If you have any doubts about the function of any Tool or Agent, you can click on its name to see a detailed description in the right pane.
For example, in the image above the com.globant.geai.web_scraper_httpx tool is selected. In the right pane you can see its name, visibility status (Public), and a brief explanation of its purpose.
To add a Tool or Agent, select the corresponding checkbox and click on the Save button.
Since April 2025 release.