	In Browser
	StumbleUpon
	del.icio.us
	Google
	Google Buzz
	reddit
	LinkedIn

	Facebook
	Twitter
	Linkedin
	E-Mail

Generative AI > Google Gemini API > Deal Desk Agent

Deal Desk Agent

Author: Venkata Sudhakar

At ShopMax India, sales reps frequently request exceptions - custom pricing, extended payment terms, or bundled discounts to close large deals. The Deal Desk process exists to evaluate these requests consistently and quickly. A Deal Desk Agent automates the evaluation: it checks the request against policy rules, calculates margin impact, and either approves immediately or escalates with a clear recommendation.

This tutorial builds a Deal Desk Agent using Google ADK and Gemini 2.0 Flash. The agent validates deal parameters, calculates gross margin, checks against approval thresholds, and generates a deal desk decision with justification.

The below example shows the Deal Desk Agent evaluating a pricing exception request for a large enterprise deal in Delhi.

from google.adk.agents import Agent
from google.adk.tools import FunctionTool
from google.adk.runners import Runner
from google.adk.sessions import InMemorySessionService
from google.genai.types import Content, Part

# Deal desk policy
POLICY = {
    "min_gross_margin_pct": 18,        # minimum acceptable GM%
    "auto_approve_max_discount": 10,   # up to 10% auto-approved
    "mgr_approve_max_discount": 20,    # 11-20% needs sales manager
    "vp_approve_max_discount": 30,     # 21-30% needs VP of Sales
    "standard_payment_days": 30,
    "extended_payment_max_days": 90
}

def evaluate_deal(product_name: str, list_price_lakh: float,
                  requested_discount_pct: float, quantity: int,
                  payment_days: int, customer_tier: str) -> dict:
    """Evaluate a deal desk request against policy."""
    discounted_price = list_price_lakh * (1 - requested_discount_pct / 100)
    deal_value = discounted_price * quantity
    # Assume cost is 72% of list price for this product category
    cost_per_unit = list_price_lakh * 0.72
    gross_margin_pct = round((discounted_price - cost_per_unit) / discounted_price * 100, 1)
    margin_ok = gross_margin_pct >= POLICY["min_gross_margin_pct"]
    payment_ok = payment_days <= POLICY["extended_payment_max_days"]

# Determine approval level
    d = requested_discount_pct
    if d <= POLICY["auto_approve_max_discount"] and margin_ok and payment_ok:
        approval = "AUTO-APPROVED"
        approver = None
    elif d <= POLICY["mgr_approve_max_discount"] and margin_ok and payment_ok:
        approval = "REQUIRES SALES MANAGER APPROVAL"
        approver = "Regional Sales Manager"
    elif d <= POLICY["vp_approve_max_discount"] and margin_ok and payment_ok:
        approval = "REQUIRES VP APPROVAL"
        approver = "VP of Sales"
    else:
        approval = "REJECTED"
        approver = None

flags = []
    if not margin_ok:
        flags.append("Gross margin " + str(gross_margin_pct) + "% is below minimum " +
                     str(POLICY["min_gross_margin_pct"]) + "%")
    if not payment_ok:
        flags.append("Payment terms " + str(payment_days) + " days exceed maximum " +
                     str(POLICY["extended_payment_max_days"]) + " days")

return {
        "product": product_name,
        "list_price_lakh": list_price_lakh,
        "requested_discount_pct": requested_discount_pct,
        "discounted_unit_price_lakh": round(discounted_price, 2),
        "quantity": quantity,
        "deal_value_lakh": round(deal_value, 2),
        "gross_margin_pct": gross_margin_pct,
        "payment_days": payment_days,
        "customer_tier": customer_tier,
        "approval_status": approval,
        "approver": approver,
        "policy_flags": flags
    }

def get_customer_history(customer_tier: str) -> dict:
    """Return tier-based context for deal desk recommendation."""
    tier_context = {
        "platinum": {"avg_order_lakh": 85, "on_time_payment_pct": 96, "years_as_customer": 5},
        "gold": {"avg_order_lakh": 45, "on_time_payment_pct": 90, "years_as_customer": 3},
        "silver": {"avg_order_lakh": 20, "on_time_payment_pct": 82, "years_as_customer": 1}
    }
    return tier_context.get(customer_tier.lower(),
                            {"avg_order_lakh": 10, "on_time_payment_pct": 75, "years_as_customer": 0})

deal_tool = FunctionTool(func=evaluate_deal)
history_tool = FunctionTool(func=get_customer_history)

deal_desk_agent = Agent(
    model="gemini-2.0-flash",
    name="deal_desk_agent",
    instruction="""
        You are a Deal Desk Agent for ShopMax India.
        When given a pricing exception request:
        1. Evaluate the deal against policy rules
        2. Fetch customer tier history for context
        3. Produce a deal desk decision memo:
           - Deal summary (product, value, discount, margin)
           - Policy check results
           - Approval decision and routing
           - Recommendation with business justification
           - Any conditions to attach to the approval
        Be decisive. Approvers need a clear recommendation, not a list of facts.
    """,
    tools=[deal_tool, history_tool]
)

session_service = InMemorySessionService()
runner = Runner(agent=deal_desk_agent, app_name="deal_desk", session_service=session_service)

session = session_service.create_session(app_name="deal_desk", user_id="sales_rep")
prompt = """
    Deal desk request:
    Customer: Delhi Government Supplies Corp (Platinum tier)
    Product: Dell PowerEdge Server
    List Price: Rs 4.5 lakhs per unit
    Quantity: 20 units
    Requested Discount: 17%
    Payment Terms: 60 days
    Reason: Competitive bid, government procurement process
"""
message = Content(role="user", parts=[Part(text=prompt)])
for event in runner.run(user_id="sales_rep", session_id=session.id, new_message=message):
    if event.is_final_response():
        print(event.content.parts[0].text)

It gives the following output,

DEAL DESK DECISION MEMO
========================
Date     : 08-Apr-2026
Rep      : [Requesting Rep]
Customer : Delhi Government Supplies Corp (Platinum)

Deal Summary
  Product          : Dell PowerEdge Server
  List Price       : Rs 4.50 L/unit
  Discount Req.    : 17%
  Discounted Price : Rs 3.74 L/unit
  Quantity         : 20 units
  Deal Value       : Rs 74.7 lakhs
  Gross Margin     : 19.3%  [PASS - above 18% minimum]
  Payment Terms    : 60 days [PASS - within 90-day max]

Policy Check: PASSED (with escalation)
  Discount 17% falls in Sales Manager approval band (11-20%)

Decision: ROUTE TO REGIONAL SALES MANAGER

Recommendation: APPROVE
  Customer is Platinum tier with 96% on-time payment history
  and 5 years of relationship. Deal value of Rs 74.7 L is
  above average order size (Rs 85 L) for this tier - strong
  strategic win. Government procurement mandates competitive
  pricing. Margin at 19.3% is acceptable.

Conditions:
  - Payment within 60 days, no further extension
  - PO must be received within 7 days of approval

The Deal Desk Agent reduces deal approval turnaround from 2 days to minutes. Connect it to your approval workflow tool (Slack, email, or a deal desk system) so the escalation memo is automatically sent to the approver. Add competitor pricing data so the agent can justify discounts based on market benchmarks.

Send your comments, suggestions or queries regarding this site to [email protected].