AI Agent Use Cases
Views —

AI Agent Use Cases

Added today

44All 189

Prompt to Prepare a Shopping Cart from a Recipe Photo

The author reports that Grok Bot receives a recipe photo, organizes the ingredients and quantities, and creates a shopping list by excluding ingredients already available at home.

Grok Bot
Life & Hobbies

1–20 of 189

Travel & BookingsWalkthrough

A Team of Agents Separately Creating Travel Research, Itinerary, and Budget Drafts

When the user enters a destination and interests, three agents handle research, itinerary creation, and budget estimation in sequence.

CrewAI

How it works

  1. The user enters the destination, number of travel days, budget in dollars, and interests. The public execution command includes --destination, --days, --budget, and --interests.
  2. After the research agent drafts information about the region and transportation, the itinerary agent receives this and organizes daily activities. The final agent reviews the previous two results and estimates the costs.
  3. The code outputs the result of the last task to the terminal. To save up to the itinerary draft as files, the code must be improved to separately store the outputs of the intermediate tasks.

Results

This is an example of a travel plan where each of the three tasks builds on the previous results. The final output is the result of the last budget estimation task.

Notes

Web search and booking tools are not connected. The place information and costs are model-generated drafts and are not based on current prices or verified business hours.

Source

GitHub Original post ↗

Example prompt · CrewAI

Set up a tool to create a draft itinerary for [destination] for [number of days]. Inputs are [interests], [companion requirements], and [total budget in dollars], and it runs in Python. Review the code for the research, itinerary, and budget agents in agents/12-travel-planner-agent, then explain the installation and execution steps.

Improve the tool so that research notes, daily itinerary, and itemized budget are each saved separately. The original does not connect to web search or booking services, so do not present model-generated prices and hours as verified values. If I provide [official place links and price data], distinguish those from estimates, and if information is missing, leave it as an item to check.

Test with inputs that include distant locations, over-budget plans, and closed days. Check that the total time and costs add up, and do not execute bookings or payments. The deliverables are the output files from the three agents, execution logs, and a human-checkable list of places and prices. Do not describe actual travel results or completed bookings.