"""
Pattern 033: Human-in-the-Loop (HITL)
Description:
Human-in-the-Loop integrates human oversight and intervention at critical
decision points in agent workflows. This pattern enables humans to approve,
correct, or guide agent actions, ensuring safety, quality, and alignment
with human values in high-stakes or ambiguous situations.
Components:
- Intervention Points: Designated moments for human input
- Approval Gates: Require human approval before proceeding
- Correction Mechanism: Allow humans to fix agent errors
- Guidance System: Humans provide direction when agent is uncertain
- Escalation Logic: Automatic escalation on confidence thresholds
- Feedback Integration: Learn from human corrections
Use Cases:
- High-stakes decision making (medical, financial, legal)
- Content moderation and review
- Quality assurance for generated content
- Training data collection and refinement
- Compliance and regulatory requirements
- Handling edge cases and ambiguity
LangChain Implementation:
Uses callback handlers and approval mechanisms to integrate human
oversight into agent workflows, with configurable intervention points
and escalation criteria.
"""
import os
from typing import List, Dict, Any, Optional, Callable
from dataclasses import dataclass, field
from datetime import datetime
from enum import Enum
from dotenv import load_dotenv
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser
from langchain_openai import ChatOpenAI
load_dotenv()
class InterventionType(Enum):
"""Types of human intervention."""
APPROVAL = "approval" # Human must approve action
CORRECTION = "correction" # Human can correct output
GUIDANCE = "guidance" # Human provides direction
REVIEW = "review" # Human reviews but doesn't block
ESCALATION = "escalation" # Automatic escalation to human
class InterventionStatus(Enum):
"""Status of intervention request."""
PENDING = "pending"
APPROVED = "approved"
REJECTED = "rejected"
CORRECTED = "corrected"
SKIPPED = "skipped"
@dataclass
class InterventionRequest:
"""Request for human intervention."""
id: str
type: InterventionType
context: str
proposed_action: str
confidence: float
timestamp: datetime = field(default_factory=datetime.now)
metadata: Dict[str, Any] = field(default_factory=dict)
# Response
status: InterventionStatus = InterventionStatus.PENDING
human_response: Optional[str] = None
correction: Optional[str] = None
response_time: Optional[datetime] = None
@dataclass
class EscalationCriteria:
"""Criteria for automatic escalation to human."""
min_confidence: float = 0.7 # Below this, escalate
sensitive_keywords: List[str] = field(default_factory=list)
high_stakes_actions: List[str] = field(default_factory=list)
require_approval_for: List[str] = field(default_factory=list)
class HumanInterface:
"""
Interface for human interaction (simulated or real).
In production, this would connect to a UI, API, or messaging system.
"""
def __init__(self, auto_approve: bool = False):
self.auto_approve = auto_approve
self.pending_requests: List[InterventionRequest] = []
def request_intervention(
self,
request: InterventionRequest
) -> InterventionRequest:
"""
Request human intervention.
In simulation mode, auto-responds. In production, would wait for human.
"""
self.pending_requests.append(request)
if self.auto_approve:
# Simulate human response
if request.type == InterventionType.APPROVAL:
request.status = InterventionStatus.APPROVED
request.human_response = "Approved by human"
elif request.type == InterventionType.CORRECTION:
# Simulate a correction
request.status = InterventionStatus.CORRECTED
request.correction = f"Corrected version: {request.proposed_action} [HUMAN EDITED]"
elif request.type == InterventionType.GUIDANCE:
request.status = InterventionStatus.APPROVED
request.human_response = "Proceed with suggested approach"
else:
request.status = InterventionStatus.APPROVED
request.response_time = datetime.now()
return request
def get_pending_count(self) -> int:
"""Get number of pending intervention requests."""
return sum(1 for r in self.pending_requests if r.status == InterventionStatus.PENDING)
def get_all_requests(self) -> List[InterventionRequest]:
"""Get all intervention requests."""
return self.pending_requests.copy()
class HITLAgent:
"""
Agent with Human-in-the-Loop integration.
Features:
- Configurable intervention points
- Automatic escalation based on criteria
- Approval gates for sensitive actions
- Learning from human feedback
"""
def __init__(
self,
human_interface: HumanInterface,
escalation_criteria: Optional[EscalationCriteria] = None,
temperature: float = 0.7
):
self.llm = ChatOpenAI(model="gpt-3.5-turbo", temperature=temperature)
self.human_interface = human_interface
self.escalation_criteria = escalation_criteria or EscalationCriteria()
self._next_request_id = 1
self.intervention_history: List[InterventionRequest] = []
# Prompt for generating response
self.response_prompt = ChatPromptTemplate.from_messages([
("system", """You are an AI assistant that works with human oversight.
Task: {task}
Generate a response and provide a confidence score (0.0-1.0) for your answer.
Respond in this format:
CONFIDENCE: [score]
RESPONSE: [your response]"""),
("user", "{task}")
])
# Prompt for confidence assessment
self.confidence_prompt = ChatPromptTemplate.from_messages([
("system", """Assess your confidence in this response on a scale of 0.0 to 1.0.
Task: {task}
Response: {response}
Consider:
- Certainty of facts
- Complexity of task
- Potential for error
- Stakes involved
Respond with just a number between 0.0 and 1.0."""),
("user", "Assess confidence")
])
def process_with_oversight(
self,
task: str,
require_approval: bool = False
) -> Dict[str, Any]:
"""
Process task with human oversight.
Args:
task: Task to process
require_approval: Force human approval regardless of confidence
Returns:
Dictionary with response and intervention details
"""
# Generate initial response
chain = self.response_prompt | self.llm | StrOutputParser()
result = chain.invoke({"task": task})
# Parse response
response, confidence = self._parse_response(result)
# Check if escalation is needed
should_escalate = (
require_approval or
self._should_escalate(task, response, confidence)
)
if should_escalate:
# Request human intervention
intervention_type = (
InterventionType.APPROVAL if require_approval or confidence tuple[str, float]:
"""Parse response and confidence from LLM output."""
lines = result.strip().split('\n')
confidence = 0.5 # Default
response = result
for line in lines:
if line.startswith("CONFIDENCE:"):
try:
conf_str = line.replace("CONFIDENCE:", "").strip()
confidence = float(conf_str)
except ValueError:
pass
elif line.startswith("RESPONSE:"):
response = line.replace("RESPONSE:", "").strip()
# If format wasn't followed, estimate confidence
if confidence == 0.5 and "CONFIDENCE:" not in result:
confidence = self._estimate_confidence(result)
return response, confidence
def _estimate_confidence(self, response: str) -> float:
"""Estimate confidence from response content."""
# Simple heuristic based on uncertainty phrases
uncertainty_phrases = [
"i think", "maybe", "possibly", "uncertain", "not sure",
"might", "could be", "perhaps", "probably"
]
response_lower = response.lower()
uncertainty_count = sum(1 for phrase in uncertainty_phrases if phrase in response_lower)
# More uncertainty phrases = lower confidence
confidence = max(0.3, 1.0 - (uncertainty_count * 0.15))
return confidence
def _should_escalate(self, task: str, response: str, confidence: float) -> bool:
"""Determine if task should be escalated to human."""
# Low confidence
if confidence Dict[str, Any]:
"""
Request human correction for a specific issue.
"""
request = InterventionRequest(
id=f"req_{self._next_request_id:04d}",
type=InterventionType.CORRECTION,
context=f"Task: {task}\nIssue: {issue_description}",
proposed_action=initial_response,
confidence=0.0, # Explicitly requesting correction
metadata={"issue": issue_description}
)
self._next_request_id += 1
# Get human correction
request = self.human_interface.request_intervention(request)
self.intervention_history.append(request)
return {
"original": initial_response,
"corrected": request.correction or initial_response,
"status": request.status.value
}
def get_statistics(self) -> Dict[str, Any]:
"""Get statistics about human interventions."""
if not self.intervention_history:
return {
"total_interventions": 0,
"by_type": {},
"by_status": {},
"avg_confidence": 0.0
}
by_type = {}
by_status = {}
for request in self.intervention_history:
# Count by type
type_key = request.type.value
by_type[type_key] = by_type.get(type_key, 0) + 1
# Count by status
status_key = request.status.value
by_status[status_key] = by_status.get(status_key, 0) + 1
avg_confidence = sum(r.confidence for r in self.intervention_history) / len(self.intervention_history)
return {
"total_interventions": len(self.intervention_history),
"by_type": by_type,
"by_status": by_status,
"avg_confidence": avg_confidence
}
def demonstrate_human_in_the_loop():
"""
Demonstrates Human-in-the-Loop with various intervention scenarios.
"""
print("=" * 80)
print("HUMAN-IN-THE-LOOP DEMONSTRATION")
print("=" * 80)
# Create human interface (simulated)
human_interface = HumanInterface(auto_approve=True)
# Configure escalation criteria
escalation_criteria = EscalationCriteria(
min_confidence=0.7,
sensitive_keywords=["medical", "financial", "legal", "delete", "confidential"],
high_stakes_actions=["approve payment", "delete data", "grant access"]
)
# Create HITL agent
agent = HITLAgent(
human_interface=human_interface,
escalation_criteria=escalation_criteria
)
# Test 1: High confidence task (no intervention)
print("\n" + "=" * 80)
print("Test 1: High Confidence Task (No Intervention Expected)")
print("=" * 80)
task1 = "What is the capital of France?"
print(f"\nTask: {task1}")
result1 = agent.process_with_oversight(task1)
print(f"Response: {result1['response']}")
print(f"Confidence: {result1['confidence']:.2f}")
print(f"HITL Used: {result1['used_hitl']}")
print(f"HITL Result: {result1['hitl_result']}")
# Test 2: Sensitive keyword (automatic escalation)
print("\n" + "=" * 80)
print("Test 2: Sensitive Content (Automatic Escalation)")
print("=" * 80)
task2 = "Should I approve this medical procedure for the patient?"
print(f"\nTask: {task2}")
result2 = agent.process_with_oversight(task2)
print(f"Response: {result2['response']}")
print(f"Confidence: {result2['confidence']:.2f}")
print(f"HITL Used: {result2['used_hitl']}")
print(f"HITL Result: {result2['hitl_result']}")
if result2['original_response']:
print(f"Original Response: {result2['original_response'][:100]}...")
# Test 3: Forced approval
print("\n" + "=" * 80)
print("Test 3: Forced Human Approval")
print("=" * 80)
task3 = "Delete all user data from the database"
print(f"\nTask: {task3}")
result3 = agent.process_with_oversight(task3, require_approval=True)
print(f"Response: {result3['response']}")
print(f"Confidence: {result3['confidence']:.2f}")
print(f"HITL Used: {result3['used_hitl']}")
print(f"HITL Result: {result3['hitl_result']}")
# Test 4: Request correction
print("\n" + "=" * 80)
print("Test 4: Human Correction Request")
print("=" * 80)
task4 = "Calculate the total revenue for Q4 2024"
initial_response = "The total revenue was $1.5M"
issue = "The calculation seems incorrect based on our records"
print(f"\nTask: {task4}")
print(f"Initial Response: {initial_response}")
print(f"Issue: {issue}")
correction_result = agent.request_correction(task4, initial_response, issue)
print(f"Corrected Response: {correction_result['corrected']}")
print(f"Status: {correction_result['status']}")
# Test 5: Low confidence (escalation)
print("\n" + "=" * 80)
print("Test 5: Low Confidence Response")
print("=" * 80)
task5 = "I'm not sure about this complex legal interpretation..."
print(f"\nTask: {task5}")
result5 = agent.process_with_oversight(task5)
print(f"Response: {result5['response']}")
print(f"Confidence: {result5['confidence']:.2f}")
print(f"HITL Used: {result5['used_hitl']}")
print(f"HITL Result: {result5['hitl_result']}")
# Show statistics
print("\n" + "=" * 80)
print("Intervention Statistics")
print("=" * 80)
stats = agent.get_statistics()
print(f"\nTotal Interventions: {stats['total_interventions']}")
print(f"By Type: {stats['by_type']}")
print(f"By Status: {stats['by_status']}")
print(f"Average Confidence: {stats['avg_confidence']:.2f}")
print(f"\nPending Requests: {human_interface.get_pending_count()}")
# Summary
print("\n" + "=" * 80)
print("SUMMARY")
print("=" * 80)
print("""
Human-in-the-Loop (HITL) provides:
✓ Human oversight at critical decision points
✓ Automatic escalation based on confidence
✓ Approval gates for sensitive actions
✓ Correction mechanisms for error fixing
✓ Guidance integration for ambiguous cases
✓ Audit trail of human interventions
This pattern excels at:
- High-stakes decision making (medical, financial, legal)
- Quality assurance and content moderation
- Handling edge cases and ambiguity
- Regulatory compliance requirements
- Training data collection
- Building trust and accountability
Intervention types:
1. APPROVAL: Human must approve before proceeding
2. CORRECTION: Human can fix errors or improve output
3. GUIDANCE: Human provides direction when uncertain
4. REVIEW: Human reviews but doesn't block execution
5. ESCALATION: Automatic routing to human based on criteria
Escalation triggers:
- Low confidence scores (< 0.7 by default)
- Sensitive keywords (medical, financial, legal, etc.)
- High-stakes actions (delete data, approve payments, etc.)
- Explicit requirement for human approval
Benefits:
- Safety: Prevents high-risk errors
- Quality: Human expertise improves outputs
- Trust: Builds confidence through oversight
- Learning: Agent learns from corrections
- Compliance: Meets regulatory requirements
- Accountability: Clear audit trail
Use HITL when you need:
- Safety in high-stakes scenarios
- Quality assurance for critical outputs
- Regulatory compliance
- Human expertise for edge cases
- Trust and transparency
- Gradual automation with human backup
""")
if __name__ == "__main__":
demonstrate_human_in_the_loop()
Agentic Design Pattern: Human In The Loop
Pattern 033: Human In The Loop – Line-by-Line Explanation
Overview
This document provides a detailed line-by-line explanation of the Human In The Loop pattern implementation.
Pattern Description
Pattern 033: Human-in-the-Loop (HITL)
Description: Human-in-the-Loop integrates human oversight and intervention at critical decision points in agent workflows. This pattern enables humans to approve, correct, or guide agent actions, ensuring safety, quality, and alignment with human values in high-stakes or ambiguous situations.
Components: – Intervention Points: Designated moments for human input – Approval Gates: Require human approval before proceeding – Correction Mechanism: Allow humans to fix agent errors – Guidance System: Humans provide direction when agent is uncertain – Escalation Logic: Automatic escalation on confidence thresholds – Feedback Integration: Learn from human corrections
Use Cases: – High-stakes decision making (medical, financial, legal) – Content moderation and review – Quality assurance for generated content – Training data collection and refinement – Compliance and regulatory requirements – Handling edge cases and ambiguity
LangChain Implementation: Uses callback handlers and approval mechanisms to integrate human oversight into agent workflows, with configurable intervention points and escalation criteria.
File Statistics
- Total Lines: 555
- Classes: 6
- Functions: 1
- Constants: 0
Module Documentation
"""
Pattern 033: Human-in-the-Loop (HITL)
Description:
Human-in-the-Loop integrates human oversight and intervention at critical
decision points in agent workflows. This pattern enables humans to approve,
correct, or guide agent actions, ensuring safety, quality, and alignment
with human values in high-stakes or ambiguous situations.
Components:
- Intervention Points: Designated moments for human input
- Approval Gates: Require human approval before proceeding
- Correction Mechanism: Allow humans to fix agent errors
- Guidance System: Humans provide direction when agent is uncertain
- Escalation Logic: Automatic escalation on confidence thresholds
- Feedback Integration: Learn from human corrections
Use Cases:
- High-stakes decision making (medical, financial, legal)
- Content moderation and review
- Quality assurance for generated content
- Training data collection and refinement
- Compliance and regulatory requirements
- Handling edge cases and ambiguity
LangChain Implementation:
Uses callback handlers and approval mechanisms to integrate human
oversight into agent workflows, with configurable intervention points
and escalation criteria.
"""
Purpose: This module implements the Human In The Loop pattern, providing a structured approach to agent design and implementation.
Import Statements
Standard Library Imports
import dataclasses.dataclass
import dataclasses.field
import datetime.datetime
import enum.Enum
import langchain_core.output_parsers.StrOutputParser
import langchain_core.prompts.ChatPromptTemplate
import os
import typing.Any
import typing.Callable
import typing.Dict
import typing.List
import typing.Optional
Purpose: Core Python functionality:
– dataclasses.dataclass: Simplified class definitions with @dataclass
– dataclasses.field: Simplified class definitions with @dataclass
– datetime.datetime: Date and time handling
– enum.Enum: Enumeration support
– langchain_core.output_parsers.StrOutputParser: Regular expression operations
– langchain_core.prompts.ChatPromptTemplate: Regular expression operations
– os: Environment variable access and system operations
– typing.Any: Type hints for better code documentation (List, Dict, Any, Optional)
– typing.Callable: Type hints for better code documentation (List, Dict, Any, Optional)
– typing.Dict: Type hints for better code documentation (List, Dict, Any, Optional)
– typing.List: Type hints for better code documentation (List, Dict, Any, Optional)
– typing.Optional: Type hints for better code documentation (List, Dict, Any, Optional)
LangChain Imports
from langchain_openai.ChatOpenAI
Purpose: LangChain framework components: – LLM Interface: Wrapper for OpenAI models – Chains: Composable building blocks for complex workflows
Other Imports
from dotenv.load_dotenv
Purpose: Additional dependencies for specialized functionality.
Classes
Class: InterventionType
Lines 46-52
Inherits from: Enum
Purpose:
Types of human intervention.
Class: InterventionStatus
Lines 55-61
Inherits from: Enum
Purpose:
Status of intervention request.
Class: InterventionRequest
Lines 65-79
Purpose:
Request for human intervention.
Class: EscalationCriteria
Lines 83-88
Purpose:
Criteria for automatic escalation to human.
Class: HumanInterface
Lines 91-138
Purpose:
Interface for human interaction (simulated or real).
In production, this would connect to a UI, API, or messaging system.
Methods:
__init__(self, auto_approve)
Lines 98-100
Purpose: Initialize the class instance with required parameters.
Parameters:
– auto_approve: Auto approve
request_intervention(self, request)
Lines 102-130
Purpose:
Request human intervention.
In simulation mode, auto-responds. In production, would wait for human.
Parameters:
– request: Request
Returns: InterventionRequest
get_pending_count(self)
Lines 132-134
Purpose:
Get number of pending intervention requests.
Parameters:
Returns: int
get_all_requests(self)
Lines 136-138
Purpose:
Get all intervention requests.
Parameters:
Returns: <ast.Subscript object at 0x000001A06D6EBA10>
Class: HITLAgent
Lines 141-390
Purpose:
Agent with Human-in-the-Loop integration.
Features:
- Configurable intervention points
- Automatic escalation based on criteria
- Approval gates for sensitive actions
- Learning from human feedback
Methods:
__init__(self, human_interface, escalation_criteria, temperature)
Lines 152-194
Purpose: Initialize the class instance with required parameters.
Parameters:
– human_interface: Human interface
– escalation_criteria: Escalation criteria
– temperature: Sampling temperature (0-1)
process_with_oversight(self, task, require_approval)
Lines 196-273
Purpose:
Process task with human oversight.
Args:
task: Task to process
require_approval: Force human approval regardless of confidence
Returns:
Dictionary with response and intervention details
Parameters:
– task: Task
– require_approval: Require approval
Returns: <ast.Subscript object at 0x000001A06D64E9D0>
_parse_response(self, result)
Lines 275-295
Purpose:
Parse response and confidence from LLM output.
Parameters:
– result: Result
Returns: <ast.Subscript object at 0x000001A06D616050>
_estimate_confidence(self, response)
Lines 297-311
Purpose:
Estimate confidence from response content.
Parameters:
– response: Response
Returns: float
_should_escalate(self, task, response, confidence)
Lines 313-330
Purpose:
Determine if task should be escalated to human.
Parameters:
– task: Task
– response: Response
– confidence: Confidence
Returns: bool
request_correction(self, task, initial_response, issue_description)
Lines 332-359
Purpose:
Request human correction for a specific issue.
Parameters:
– task: Task
– initial_response: Initial response
– issue_description: Issue description
Returns: <ast.Subscript object at 0x000001A06D159850>
get_statistics(self)
Lines 361-390
Purpose:
Get statistics about human interventions.
Parameters:
Returns: <ast.Subscript object at 0x000001A06D70FE10>
Functions
demonstrate_human_in_the_loop()
Lines 393-550
Purpose:
Demonstrates Human-in-the-Loop with various intervention scenarios.
Key Concepts Summary
Pattern Overview
The Human In The Loop pattern provides a structured approach to building AI agents with specific capabilities and behaviors. This implementation leverages LangChain’s composable components to create a robust and maintainable solution.
Core Components
- Agent Class: Main implementation of the pattern logic
- Prompt Engineering: Carefully crafted prompts to guide LLM behavior
- Tool Integration: Optional tools for extended functionality
- Memory Management: State handling across interactions
- Output Parsing: Structured response processing
When to Use
✅ Good for: – Scenarios requiring this specific pattern’s strengths – Applications needing structured agent behavior – Systems where explainability is important – Production environments requiring reliability
❌ Not ideal for: – Simple single-prompt tasks – Scenarios requiring different pattern characteristics – Resource-constrained environments (if computationally intensive)
Best Practices Demonstrated
1. Environment Configuration
- ✅ Use
.envfiles for API keys - ✅ Never hardcode credentials
- ✅ Validate environment variables on startup
2. Error Handling
- ✅ Graceful degradation on failures
- ✅ Informative error messages
- ✅ Logging for debugging
3. Prompt Engineering
- ✅ Clear, specific instructions
- ✅ Structured output formats
- ✅ Few-shot examples when beneficial
4. Code Organization
- ✅ Separation of concerns (class-based design)
- ✅ Type hints for clarity
- ✅ Docstrings for documentation
5. Performance Optimization
- ✅ Appropriate temperature settings
- ✅ Token budget management
- ✅ Caching where applicable
LangChain Components Used
- ChatOpenAI: OpenAI model interface
- Chains: Sequential component composition
Why LangChain? – Composable abstractions for LLM applications – Rich ecosystem of integrations – Production-ready patterns – Active community and development
Execution Flow
1. Initialize agent with configuration
↓
2. Load environment and dependencies
↓
3. Prepare prompt with input data
↓
4. Invoke LLM through LangChain
↓
5. Parse and validate response
↓
6. Return formatted result
Key Decision Points: – Model selection (temperature, model type) – Prompt structure (few-shot vs zero-shot) – Error handling strategy – Output format validation
Performance Considerations
Computational Cost
- LLM Calls: Primary cost factor (based on tokens)
- Latency: Depends on model and input size
- Memory: Minimal for most patterns
Optimization Strategies
- Prompt Optimization: Minimize tokens while maintaining clarity
- Caching: Cache repeated queries/responses
- Model Selection: Use smaller models when appropriate
- Batch Processing: Group similar requests
- Streaming: For long responses, stream tokens
Cost Estimates (Approximate)
- GPT-3.5-turbo: ~$0.002 per request
- GPT-4: ~$0.03 per request
- GPT-4-turbo: ~$0.01 per request
Actual costs vary based on input/output token counts
Common Pitfalls to Avoid
1. Prompt Design Issues
❌ Vague or ambiguous instructions
✅ Clear, specific, testable prompts
2. Error Handling
❌ Ignoring API errors or rate limits
✅ Implement retry logic and graceful degradation
3. Token Management
❌ Exceeding model token limits
✅ Truncate inputs, chunk processing
4. Security
❌ Exposing API keys in code
✅ Use environment variables
5. Testing
❌ Only testing happy paths
✅ Test edge cases, failures, malformed inputs
6. Cost Management
❌ Unmonitored API usage
✅ Implement budgets, alerts, and monitoring
Further Improvements
Production Enhancements
- Monitoring: Add logging, metrics, tracing
- Validation: Input/output schema validation
- Rate Limiting: Implement request throttling
- Caching: Redis or in-memory cache for responses
- Async Support: Non-blocking operations for scalability
Feature Extensions
- Multi-Model Support: Switch between providers
- Fallback Models: Graceful degradation strategy
- Custom Tools: Domain-specific function calling
- Advanced Memory: Vector stores for long-term memory
- Evaluation: Automated testing and benchmarking
Code Quality
- Testing: Unit tests, integration tests
- Documentation: API docs, usage examples
- Type Safety: Stricter type hints
- Configuration: Externalized config files
Related Patterns
- Pattern 032
- Pattern 034
Pattern Families: – Reasoning Patterns (001-011) – Multi-Agent Patterns (012-019) – Tool Use Patterns (020-025) – Memory Patterns (026-031) – And many more…
Summary
The Human In The Loop pattern provides a powerful approach to building AI agents with specific capabilities. This implementation demonstrates:
✅ Key Strengths: – Clean, maintainable code structure – LangChain integration for composability – Type hints and documentation – Production-ready patterns
🎯 Best Use Cases: – Applications requiring this pattern’s characteristics – Systems needing structured agent behavior – Production environments requiring reliability
📚 Learning Resources: – LangChain Documentation: https://python.langchain.com/ – Pattern Documentation: See related patterns – OpenAI API Reference: https://platform.openai.com/docs/
File: 033_human_in_the_loop.py
Total Lines: 555
Pattern Category: AI Agent Design Pattern #033
Generated from actual implementation code
For educational and reference purposes
Leave a Reply