Trip Planner
A hierarchical multi-agent system where a manager agent coordinates six specialized sub-agents to produce a complete, budget-aware travel itinerary — with charts and a downloadable PDF — from a single prompt.
- Agents — 1 Manager + 6 Sub-Agents
- Output Formats
- Prompt to Full Itinerary
ArchitectureCrewAI · HierarchicalBuilt WithCrewAI · Multi-Agent · Gradio · LLMs · ReportLab · matplotlib · plotly · Python
The Problem
Planning a trip that respects a budget is hard, slow, and fragmented
A good trip plan requires several separate skills working in concert — researching what a destination actually offers, identifying the right activities for the traveler's preferences, allocating a finite budget across transport, accommodation, activities, and food without overspending, and finally writing a coherent day-by-day itinerary that ties it all together.
Doing this manually takes hours of cross-referencing across search engines, booking platforms, and spreadsheets — and even then the result is rarely budget-optimal or well-structured.
The Solution
A manager agent that delegates, coordinates, and delivers
Trip Planner Agent is a seven-agent hierarchical CrewAI system where a manager agent orchestrates six specialized sub-agents — a budget analyst, destination researcher, transport specialist, accommodation specialist, local gems specialist, and itinerary builder — each owning one stage of the planning process.
The user specifies destination, budget, duration, and preferences through a Gradio UI. The manager coordinates the sub-agents into a coherent plan and the system delivers three outputs: a structured day-by-day itinerary, a budget breakdown chart, and a downloadable PDF report — all from a single prompt.
Key Outcome
From a single prompt, the system produces a complete, budget-compliant travel itinerary — with a budget breakdown chart and a professionally formatted downloadable PDF — by delegating the full planning workflow to a coordinated team of six specialized agents.
Technical Deep Dive
Architecture & Design
Agent Hierarchy
Stage by Stage
- Manager Agent
Hierarchical Coordinator
Orchestrates the full planning workflow by delegating tasks to sub-agents, monitoring their outputs, and synthesising the results into a coherent plan. The manager never executes research or writing directly — it coordinates and guides the specialist agents toward the final deliverable.
- Sub-Agent 1
Budget Analyst
Allocates the user’s total budget across six spending categories — flights, accommodation, food & dining, activities, local transport, and miscellaneous — and converts all figures to the destination currency. Runs first so every downstream agent receives hard per-category caps before beginning its own research. Produces the cost breakdown that feeds directly into the chart visualisations and PDF report.
- Sub-Agent 2
Destination Researcher
Uses real-time web search tools to build a structured destination profile — covering visa and entry requirements, best neighbourhoods by travel style, local transport options, cultural etiquette, safety tips, and current travel advisories. Provides the factual foundation that all downstream agents build on.
- Sub-Agent 3
Transport Specialist
Searches for flight options across budget, mid-range, and premium tiers. Evaluates airlines, journey times, and layover options — recommending the combination that fits within the Budget Analyst’s flights allocation and the traveller’s style preference.
- Sub-Agent 4
Accommodation Specialist
Identifies accommodation options that match the traveller’s style preference — from hostels to boutique hotels — while enforcing a per-night budget cap derived from the Budget Analyst’s allocation. Produces a shortlist with nightly rate, location, standout amenities, and booking platform recommendation for the Itinerary Builder.
- Sub-Agent 5
Local Gems Specialist
Researches activities, restaurants, and off-the-beaten-path experiences tailored to the user’s stated interests and dietary preferences. Goes beyond generic tourist highlights to surface local picks and hidden gems that fit the trip’s budget, duration, and traveller style.
- Sub-Agent 6
Itinerary Builder
Synthesises all upstream outputs — destination research, flight options, accommodation shortlist, budget allocations, and local recommendations — into a structured, readable day-by-day itinerary. Produces the final narrative rendered in the Gradio UI and exported to the downloadable PDF report.
Why It Works
Key Design Decisions
Hierarchical architecture separates coordination from execution
The manager agent never performs research, planning, or writing directly. Its sole responsibility is orchestration — deciding which sub-agent to engage, in what order, and how to synthesise their outputs. This clean separation keeps each agent focused on a single well-defined task and makes the system easier to extend with new specialist agents.
Budget constraint enforced at the agent level
Rather than applying budget limits as a post-processing filter, the Budget Analyst agent runs first and actively allocates costs across six categories — flights, accommodation, food & dining, activities, local transport, and miscellaneous — before any other agent begins. This means flight options, accommodation shortlists, and activity picks are all built around realistic cost constraints from the start — not retrofitted to a budget after the fact.
Three output formats serve different user needs
The system produces three distinct outputs from the same planning run — a structured itinerary for reading, an interactive chart for budget visualisation, and a downloadable PDF for sharing and offline use. Each format serves a different consumption pattern without requiring the user to run the pipeline multiple times.
Results & Metrics
What the system delivers
- Specialized Agents
- 1 manager coordinating 6 sub-agents — each owning one stage of the planning process
- Output Formats
- Structured itinerary, budget breakdown chart, and downloadable PDF — all from one run
- Prompt to Full Plan
- Destination, budget, duration, and preferences — one input drives the entire pipeline
Budget-compliant itineraries
The Budget Analyst agent allocates costs across six categories during planning — not as a post-processing filter. Every itinerary produced stays within the user’s defined budget constraints from the first agent call to the final PDF.
Budget breakdown chart
Budget allocation across six categories — flights, accommodation, food & dining, activities, local transport, and miscellaneous — is rendered as a horizontal bar chart in the Gradio UI, giving users an immediate visual breakdown of where their budget is going before downloading the full report.
Downloadable PDF report
The full trip plan — itinerary, budget details, and activity summaries — is exported as a professionally formatted PDF via ReportLab. The report is shareable and readable offline without requiring access to the Gradio interface.
Real-time destination research
The Destination Researcher agent uses live web search tools to pull current information on destinations, activities, and pricing — ensuring the plan reflects real-world conditions rather than relying on static training data.
Agent Roster
| Agent | Responsibility |
|---|---|
| Manager Agent | Orchestrates sub-agents, allocates tasks, synthesises final plan |
| Budget Analyst | Allocates budget across 6 categories, converts currency for international trips — runs first so all other agents receive firm cost caps |
| Destination Researcher | Real-time web search for destination profile, highlights, and travel tips |
| Transport Specialist | Finds flights across budget/mid/premium tiers, recommends optimal transport options |
| Accommodation Specialist | Matches hotels/hostels to travel style and enforces per-night budget cap |
| Local Gems Specialist | Curates activities and dining by interest, surfaces hidden local picks by dietary preference |
| Itinerary Builder | Synthesises all outputs into a structured, readable day-by-day trip plan |
Tools
Tech Stack
| Technology | Purpose |
|---|---|
| CrewAI | Hierarchical multi-agent orchestration with manager and sub-agents |
| OpenAI API | LLM backbone for agent reasoning, research, and content generation |
| Gradio | Web UI for trip parameter input and multi-format output display |
| matplotlib | Budget breakdown chart generation |
| ReportLab | PDF report generation for downloadable trip plan |
| Web Search Tools | Real-time destination, activity, and pricing research by agents |
| Python | Core language and pipeline orchestration |