Input Defaults & UI Patterns in watsonx Orchestrate: Supercharging UX with Smart Pydantic Schemas
How to guide users, reduce manual input errors, and auto-populate forms using Pydantic Field schemas.
Why This Pattern Matters
In enterprise workflows like financial approvals or travel requests, users often don't know the exact parameters required, leading to repetitive validation errors. By enforcing smart defaults and range constraints with Pydantic in watsonx Orchestrate, our AI agent can pre-fill sensible defaults (e.g. $1,000 standard budget) and let users override only what is necessary.
Step 1: Write Tool Logic (financial_tool.py)
from ibm_watsonx_orchestrate import tool
from pydantic import BaseModel, Field
class FinancialApprovalInput(BaseModel):
"""Input schema for financial approval with UI defaults."""
amount: float = Field(
default=1000.0,
description="The amount to approve (USD)",
ge=0.0,
le=1000000.0
)
reason: str = Field(
default="Standard equipment refresh",
description="Reason for the approval request"
)
urgent: bool = Field(
default=False,
description="Mark as urgent for expedited manager review"
)
@tool
def request_financial_approval(input: FinancialApprovalInput) -> str:
"""Submit a financial approval request with pre-filled defaults."""
urgency = "🚨 URGENT" if input.urgent else "📋 Standard"
return f"""
{urgency} Financial Approval Request Submitted
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Amount: ${input.amount:,.2f} USD
Reason: {input.reason}
Status: Pending Department Review
Tracking: WXO-REQ-{(hash(input.reason) % 100000):05d}
"""Step 2: Dependencies (requirements.txt)
ibm-watsonx-orchestrate>=2.9.0
pydantic>=2.0.0Step 3: Agent Manifest (agent.yaml)
spec_version: v1
kind: native
name: finance_approval_agent
title: "Finance Approval Assistant"
description: "Assists employees with submitting financial purchase requests with smart defaults."
model: ibm/granite-3-8b-instruct
instructions: |
You are an enterprise finance assistant.
When a user requests financial approval, inspect their prompt for amounts and reasons.
If they do not specify an amount, use the tool's smart default.
Always confirm the submission details back to the user.
tools:
- request_financial_approvalStep 4: Live Execution
orchestrate chat ask -n finance_approval_agent "I need financial approval for new test monitors for $3500"╭─ 🤖 finance_approval_agent ───────────────────────────────────────────╮
│ │
│ 📋 Standard Financial Approval Request Submitted │
│ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ │
│ Amount: $3,500.00 USD │
│ Reason: new test monitors │
│ Status: Pending Department Review │
│ Tracking: WXO-REQ-84729 │
│ │
│ Your request has been submitted to the finance desk for processing. │
╰────────────────────────────────────────────────────────────────────────╯💡 Pro Tips & Architecture Insights
💡 Pydantic v1 vs v2 compatibility
Ensure you use standard Pydantic Field definitions with ge/le constraints. WxO ADK 2.9+ bundles Pydantic v2.
💡 Default UI Rendering
When an agent invokes a tool in web chat with defaults, the parameters are pre-populated, dramatically decreasing conversational friction.