Financial Advisor

Multi-agent financial advisory system with data analysis, trading strategies, execution planning, and risk evaluation.

Architecture

Root coordinator delegates to 4 specialist sub-agents via AgentTool:

• data_analyst_agent → market_data_analysis_output

• trading_analyst_agent → proposed_trading_strategies_output

• execution_analyst_agent → execution_plan_output

• risk_analyst_agent → final_risk_assessment_output

Native ADK

Original uses 8+ files across 5 directories:

financial_advisor/
├── __init__.py
├── agent.py
├── prompt.py
└── sub_agents/
    ├── data_analyst/
    │   ├── __init__.py
    │   ├── agent.py
    │   └── prompt.py
    ├── trading_analyst/
    │   ├── __init__.py
    │   ├── agent.py
    │   └── prompt.py
    ├── execution_analyst/
    │   ├── __init__.py
    │   ├── agent.py
    │   └── prompt.py
    └── risk_analyst/
        ├── __init__.py
        ├── agent.py
        └── prompt.py

sub_agents/data_analyst/agent.py (click to expand)

from google.adk import Agent
from google.adk.tools import google_search
from . import prompt
MODEL = "gemini-2.5-pro"
data_analyst_agent = Agent(
    model=MODEL, name="data_analyst_agent",
    instruction=prompt.DATA_ANALYST_PROMPT,
    output_key="market_data_analysis_output",
    tools=[google_search],
)

sub_agents/trading_analyst/agent.py (click to expand)

from google.adk import Agent
from . import prompt
MODEL = "gemini-2.5-pro"
trading_analyst_agent = Agent(
    model=MODEL, name="trading_analyst_agent",
    instruction=prompt.TRADING_ANALYST_PROMPT,
    output_key="proposed_trading_strategies_output",
)

sub_agents/execution_analyst/agent.py (click to expand)

from google.adk import Agent
from . import prompt
MODEL = "gemini-2.5-pro"
execution_analyst_agent = Agent(
    model=MODEL, name="execution_analyst_agent",
    instruction=prompt.EXECUTION_ANALYST_PROMPT,
    output_key="execution_plan_output",
)

sub_agents/risk_analyst/agent.py (click to expand)

from google.adk import Agent
from . import prompt
MODEL = "gemini-2.5-pro"
risk_analyst_agent = Agent(
    model=MODEL, name="risk_analyst_agent",
    instruction=prompt.RISK_ANALYST_PROMPT,
    output_key="final_risk_assessment_output",
)

# financial_advisor/agent.py
from google.adk.agents import LlmAgent
from google.adk.tools.agent_tool import AgentTool
from . import prompt
from .sub_agents.data_analyst import data_analyst_agent
from .sub_agents.execution_analyst import execution_analyst_agent
from .sub_agents.risk_analyst import risk_analyst_agent
from .sub_agents.trading_analyst import trading_analyst_agent

MODEL = "gemini-2.5-pro"

financial_coordinator = LlmAgent(
    name="financial_coordinator",
    model=MODEL,
    description=(
        "guide users through a structured process to receive financial "
        "advice by orchestrating a series of expert subagents..."
    ),
    instruction=prompt.FINANCIAL_COORDINATOR_PROMPT,
    output_key="financial_coordinator_output",
    tools=[
        AgentTool(agent=data_analyst_agent),
        AgentTool(agent=trading_analyst_agent),
        AgentTool(agent=execution_analyst_agent),
        AgentTool(agent=risk_analyst_agent),
    ],
)
root_agent = financial_coordinator

Fluent API

2 files, flat directory:

financial_advisor/
├── __init__.py
├── agent.py
└── prompt.py

# agent.py
from adk_fluent import Agent
from dotenv import load_dotenv
from google.adk.tools import google_search

from .prompt import (
    DATA_ANALYST_PROMPT,
    EXECUTION_ANALYST_PROMPT,
    FINANCIAL_COORDINATOR_PROMPT,
    RISK_ANALYST_PROMPT,
    TRADING_ANALYST_PROMPT,
)

load_dotenv()

MODEL = "gemini-2.5-pro"

data_analyst = (
    Agent("data_analyst_agent", MODEL)
    .instruct(DATA_ANALYST_PROMPT)
    .tool(google_search)
    .writes("market_data_analysis_output")
)

trading_analyst = (
    Agent("trading_analyst_agent", MODEL)
    .instruct(TRADING_ANALYST_PROMPT)
    .writes("proposed_trading_strategies_output")
)

execution_analyst = (
    Agent("execution_analyst_agent", MODEL)
    .instruct(EXECUTION_ANALYST_PROMPT)
    .writes("execution_plan_output")
)

risk_analyst = (
    Agent("risk_analyst_agent", MODEL)
    .instruct(RISK_ANALYST_PROMPT)
    .writes("final_risk_assessment_output")
)

root_agent = (
    Agent("financial_coordinator", MODEL)
    .describe(
        "guide users through a structured process to receive financial "
        "advice by orchestrating a series of expert subagents"
    )
    .instruct(FINANCIAL_COORDINATOR_PROMPT)
    .writes("financial_coordinator_output")
    .agent_tool(data_analyst)
    .agent_tool(trading_analyst)
    .agent_tool(execution_analyst)
    .agent_tool(risk_analyst)
    .build()
)

What Changed

• 4x AgentTool(agent=...) → .agent_tool()

• output_key= → .writes()

• instruction= → .instruct()

• description= → .describe()

• 8+ files across 5 directories → 2 files in 1 directory

• No __init__.py re-export chain needed

• No separate sub-agent packages

Metrics

Metric	Native	Fluent	Reduction
Agent definition files	5	1	80%
Total Python files	10	3	70%
Directories	5	1	80%
import statements	15+	5	67%