How to Use a Tool in AI Agents?

Tools are a core capability within the GPTBots ecosystem, allowing AI Agents and Workflows to evolve beyond simple text generation. By utilizing tools, Agents can fetch real-time data, trigger specific actions, and interact with external systems to solve real-world business challenges.

This guide explores the practical implementation of tools across different Agent architectures and structured Workflows.

The Agent Configuration Architecture

The following diagram illustrates how "Tools" serve as a critical component in the GPTBots DevSpace, integrated alongside the LLM and Knowledge Base.

---
config:
  layout: dagre
  look: handDrawn
---
flowchart TB
    n2["Agent"] --> n6["Configuration"]
    n6 --> n12["LLM"] & n13["Knowledge Base"] & n14["Database"] & n15["Workflows"] & n16["Tools"] & n17["Memory"] & n18["Welcome Guide"] & n19["Human handoff"] & n20["Input / Output"] & n21["Safety Mechanism"]

    n2@{ shape: rect}
    style n2 color:#000000
    style n16 color:#D50000

1. Implementing Tools in GPTBots Agents

GPTBots offers three distinct Agent types: Standard Agent, Flow Agent, and Multi-Agent. While each type supports tool integration, the configuration process is tailored to the specific logic of the agent.

1.1 Tools in a Standard Agent

For a standard Agent, tools are configured globally within the Agent’s main settings:

1. Open the Agent Configuration page.
2. Locate the Tools section.
3. Click Add, search for the desired tool, and integrate it.

Operational Logic: Once added, the Agent uses its internal reasoning to automatically decide whether to invoke a tool based on the user's input and conversation context.

1.2 Tools in a Flow Agent

In a Flow Agent, tools are scoped to the LLM component within the visual canvas rather than a global configuration:

1. Open the Flow Agent editor and click Edit.
2. Select the specific LLM component node.
3. Locate the Tools section within that node's settings and add your tools.

Operational Logic: Tool usage is specifically governed by the requirements of that specific LLM node within the larger flow logic.

1.3 Tools in a Multi-Agent System

For Multi-Agent orchestration, tools are assigned to individual sub-agents to maintain specialized roles:

1. Navigate to the Agent section within the Multi-Agent configuration.
2. Select the specific Agent that requires the capability.
3. Locate and configure the Tools section for that agent.

Operational Logic: This allows for collaborative behaviors where different agents within one system possess unique, specialized toolsets.

2. Leveraging Tools within AI Workflows

Beyond Agents, tools are essential building blocks for GPTBots Workflows, offering two distinct methods of application depending on the level of automation required.

2.1 Integrating Tools in the LLM Component

One common method is to empower an LLM node within a workflow to use tools autonomously:

1. Select the LLM component in your Workflow canvas.
2. Access the Tools section in the settings panel.
3. Add the tools you wish to make available.

Operational Logic: During workflow execution, the LLM determines if a tool call is necessary to process the current step.

2.2 Using a Tool as a Standalone Workflow Component

Alternatively, you can deploy a tool as an explicit, deterministic step in your workflow:

1. Locate the Tools component in the Workflow builder library.
2. Drag and drop the specific tool into the canvas.
3. Connect the tool node to your logic flow.

Operational Logic: In this scenario, the tool acts as a fixed node. This makes the execution path fully controllable and transparent, as the tool is triggered as a mandatory step in the sequence.

Conclusion

Tools are the primary mechanism for extending the utility of GPTBots. Whether you are configuring autonomous Agents or building deterministic Workflows, tools enable your AI to break out of the chat box and perform meaningful actions across your enterprise's external systems.