Conditional Routing

Real agents rarely follow one path. Routing lets a single graph handle billing vs. technical vs. escalation flows without maintaining separate agent binaries.

Conditional routing is a airport security checkpoint: same entrance, but travelers go to domestic, international, or additional screening based on their ticket.

Scenario

After `execute_sql`, router checks `state['row_count']`: 0 → `broaden_query`, >100 → `summarize_only`, else → `format_table`.

Solution

In Agent Frameworks, apply Conditional Routing to this scenario: After `execute_sql`, router checks `state['row_count']`: 0 → `broaden_query`, >100 → `summarize_only`, else → `format_table`. Identify the inputs, run the technique, validate the output, and note one thing you would monitor in production.

Do this before moving to the next module — reading alone is not enough.

Open the Code Walkthrough below and run it locally. Change one parameter related to Conditional Routing (e.g. model, temperature, top_k, or tool name), observe the difference in output, and write 2–3 sentences explaining what changed.

Highlighted lines show where Conditional Routing happens in the code.

1def sql_router(state):  # define a reusable function2    if state.get("sql_error"):3        return "fix_sql"  # return the result4    if state["row_count"] > 100:5        return "summarize"  # return the result6    return "format"  # return the result7
8graph.add_conditional_edges("execute_sql", sql_router,  # key line for Conditional Routing9    {"fix_sql": "fix_sql", "summarize": "summarize", "format": "format"})

• Treating Conditional Routing as a black box without evaluation
• Ignoring cost and latency in production
• Skipping error handling for langgraph conditional routing

Quick recap — the most important points from this module.