Skip to content
Nappai Documentation

Agent Router

Agent Router is a smart navigation tool within the Nappai automation system. Think of it as a traffic director or a receptionist for your data workflows. When you ask a question or submit a task, this component analyzes what you need and automatically sends it to the right specialized agent or process. Instead of manually deciding which tool to use, Agent Router reads your input and decides the best path to take, ensuring your request gets handled efficiently.

How it Works

Internally, Agent Router acts as a decision-making node in your automation flow. It uses artificial intelligence to understand the intent behind your text or data.

  1. Analysis: It looks at the content of your input (e.g., a search query or a data request).
  2. Decision: It compares this content against available routes in your system.
  3. Routing: It passes the information to the specific agent or service best equipped to handle it (e.g., sending a “find price” request to a search agent or a “calculate total” request to a math agent).

It does not perform the final task itself; rather, it ensures the task is handed off to the right place. Note that this specific version is marked as a “legacy” component, meaning it is a stable, older version of the router. While it works reliably, newer versions may offer updated features.

Connection & Credentials

This component does not require external API credentials or complex configuration panels. It is ready to use directly within your Nappai workflow once placed.

Operations

This component does not have selectable operations. It automatically performs its routing function based on the input provided.

Inputs

This component does not require manual input configuration in the dashboard. It automatically receives data from previous steps in your workflow. The system handles the data flow, so you do not need to define specific fields like “Query” or “Input Text” here.

Outputs

The component produces a decision signal that directs the workflow to the next step. This output is typically a status or a pointer indicating which agent or process should handle the next part of the task. You can connect this output to subsequent nodes to continue your automation chain.

Output Data Example (JSON)

json { “route_decision”: “search_agent”, “status”: “completed”, “next_hop”: “data_search_node” }

Connectivity

Typical Connections:

  • From Previous Nodes: Connects to any component that generates text, queries, or data requests. It is usually placed early in a workflow to sort incoming tasks.
  • To Specialized Agents: Connects to specific “worker” agents (like Search, Database, or Calculation agents) based on the type of task.

Why connect here? Placing Agent Router at the start of a multi-step process ensures that different types of requests (e.g., “Find info” vs. “Summarize data”) are sent to the most appropriate tools, preventing errors and improving speed.

Usage Example

Scenario: You are building a workflow that answers customer support questions.

  1. Input: A customer types, “What is the return policy for electronics?”
  2. Agent Router: Analyzes the text and recognizes it as a “policy” question.
  3. Routing: Directs the request to the “Legal/Policy Agent” instead of the “Sales Agent.”
  4. Result: The Policy Agent retrieves the correct information and returns it to you.

Without Agent Router, you might have to manually choose which agent to call for every single query. With it, the routing happens automatically.

Tips and Best Practices

  • Use for Mixed Inputs: If your workflow handles various types of requests (e.g., both searching data and calculating numbers), always place Agent Router at the start to sort them.
  • Legacy Status: Be aware that this is a legacy component. If you are starting a brand new workflow from scratch, check if a newer, non-legacy Router version is available for potentially better performance.
  • Simple Placement: You do not need to configure inputs. Just plug your data source into this node and the router into your specific agents.

Security Considerations

As an AI-driven component, ensure that the data you send through the Agent Router does not contain sensitive personal information if your downstream agents have limited security protocols. The router itself does not store data; it simply passes it along.