Lab 21.6 Solution: Market Analyst with Deterministic Edges

Goal

This file contains the complete code for the agent.py script using the ADK 2.0 Deterministic Workflow pattern.

market_analyst/agent.py

from __future__ import annotations
from pydantic import BaseModel
from google.adk import Agent, Workflow
from typing import Literal

# 1. Define the Classification Schema
# Using Pydantic is CRITICAL here because the Workflow router 
# matches the node's output against the dictionary keys.
class MarketRoute(BaseModel):
    currency: Literal["USD", "EUR", "GBP"]

# 2. Create the Classifier Node
classifier = Agent(
    name="classifier",
    model="gemini-3.5-flash",
    instruction="Extract the currency (USD, EUR, or GBP) from the user's request. Return ONLY the JSON.",
    output_schema=MarketRoute
)

# 3. Create Specialist Agents (Nodes)
usd_analyst = Agent(
    name="usd_analyst",
    model="gemini-3.5-flash",
    instruction="Provide a brief, bullish outlook for the US Dollar."
)

eur_analyst = Agent(
    name="eur_analyst",
    model="gemini-3.5-flash",
    instruction="Provide a brief, cautious outlook for the Euro."
)

gbp_analyst = Agent(
    name="gbp_analyst",
    model="gbp_analyst", # Tip: Use name as a string or the variable
    model="gemini-3.5-flash",
    instruction="Provide a brief, neutral outlook for the British Pound."
)

# 4. Build the Deterministic Workflow
# The 'edges' list defines the explicit structure of the graph.
root_agent = Workflow(
    name="MarketSystem",
    edges=[
        # Rule 1: The workflow always starts by running the classifier.
        ("START", classifier),
        
        # Rule 2: Route the output of the classifier to a specialist.
        # ADK 2.0 automatically looks at the 'currency' field of the 
        # MarketRoute object returned by the classifier.
        (classifier, {
            "USD": usd_analyst,
            "EUR": eur_analyst,
            "GBP": gbp_analyst
        })
    ]
)

Self-Reflection Answers

1. What happens if the classifier returns a value that isn't in your dictionary (e.g., "JPY")?

   • Answer: The workflow will fail with a routing error because the graph engine doesn't know where to go next. This is why using Literal in Pydantic is so important—it forces the LLM to choose from your defined paths.

2. Can you add an "other" key to the dictionary to handle unknown inputs?

   • Answer: Yes! You could update your MarketRoute schema to include "OTHER" and then add a corresponding entry in the edges dictionary to point to a general-purpose agent.

3. How does this approach compare to the @node dynamic workflow in terms of code complexity?

   • Answer: The deterministic Workflow is much more concise for simple routing. You don't have to write async def, handle Context, or manually call await ctx.run_node(). The framework handles all the plumbing for you. However, it is less flexible if you need complex Python logic (like loops) between nodes.