Building Trustworthy AI Agents with Human-in-the-Loop Approval

Human-in-the-Loop Travel Booking Agent

The Human-in-the-Loop Travel Booking Agent system, developed by Michal Sutter, enables users to review and approve travel plans before execution. This system uses LangGraph and Streamlit to create a transparent and controllable workflow.

Why This Matters

In reality, AI systems often lack transparency and accountability, leading to unintended consequences. Ideal models assume perfect autonomy, but human-in-the-loop approval ensures that AI agents act with human oversight, preventing potential failures and costs. For instance, a travel booking agent without human approval could lead to incorrect bookings or financial losses.

Key Insights

• LangGraph interrupts enable human approval in AI workflows (Source: LangGraph documentation)
• Streamlit supports the creation of interactive, web-based interfaces for AI applications (Example: Travel booking agent interface)
• OpenAI models can be used for generating travel plans, but require validation and approval (Tool: OpenAI, User: Michal Sutter)

Working Examples

Setup and installation of required libraries and utilities

!pip -q install -U langgraph openai streamlit pydantic
!npm -q install -g localtunnel
import os, getpass, textwrap, json, uuid, time
if not os.environ.get("OPENAI_API_KEY"):
    os.environ["OPENAI_API_KEY"] = getpass.getpass("OPENAI_API_KEY (hidden input): ")
os.environ.setdefault("OPENAI_MODEL", "gpt-4.1-mini")

Practical Applications

• Company: Travel booking agency, Behavior: Human-in-the-loop approval for travel plans
• Pitfall: Lack of transparency in AI decision-making, Consequence: Incorrect bookings or financial losses

References:

• https://www.marktechpost.com/2026/02/16/how-to-build-human-in-the-loop-plan-and-execute-ai-agents-with-explicit-user-approval-using-langgraph-and-streamlit/