Make AI Agents
AI agent is an autonomous system that acts within the defined context and constraints to achieve a goal. To achieve the goal, the AI agent uses reasoning powered by a large language model (LLM). The AI agent is like your assistant who performs tasks for you and uses its best judgment to complete them. This makes the agent flexible, but it might be also harder to predict the agent's results.
Using agents changes the way you approach automation. Instead of thinking about how to reach a goal, and which modules and transformations you need to use, you can now focus on just defining the goal. With the AI agent, your automation design becomes goal-oriented instead of process-oriented.
When you create an AI agent, you provide the agent with its definition, purpose, and constraints in the agent's description. These definitions form the base of how your agent operates. When you use an agent in a , you provide the agent with tools to reach the goal you send in the message. The agent's tools are that the agent runs on your behalf to complete the task you requested. The agent uses name and description to determine when to use them to achieve the goal.
 doesn't add blueprints to the AI agent's context. All agent's reasoning happens based on the information in the agents description and its tools () names and descriptions. doesn't share data anywhere.
Providing the right agent with the right tools can reduce the complexity of your automations. The agent can use reasoning to decide which tools to use depending on the inputs, which can substitute for having multiple routes and filters.
If you want to send data to the agent, you have to use outputs and the Scenarios > Return output module.
In the next sections, we will build an AI agent to manage inventory of a small shop to showcase the concepts behind the AI agent. Further on, we will look at the interfaces to manage your AI agents and mention a full reference of the AI Agents app modules.