📖 Project Story

Inspiration

TripCraft started from a very personal frustration.

We are a couple of international undergraduate computer science students studying in the United States. We deeply value the time we have here, and during every academic break, we travel and explore different places. Traveling represents freedom, discovery, and a break from academic pressure—something we truly look forward to.

Ironically, planning those trips often became the most stressful part.

Before each trip, we found ourselves spending hours and even days jumping between flight search engines, hotel platforms, maps, blogs, and spreadsheets. We also debated over trade-offs: speed versus cost, comfort versus distance, ambition versus exhaustion. What should have been exciting sometimes turned into frustration, simply because there were too many decisions and too little clarity about what was actually feasible.

At some point, we realized that the mental load of planning was actively draining our joy for traveling.

Instead of asking how to plan better, we asked a different question:

What if we could delegate the hardest travel-planning decisions to AI—without losing realism, control, or trust?

That question became the foundation of TripCraft.

What it does

TripCraft is an AI-powered travel planning assistant that turns vague ideas—or even just a travel “vibe”—into a fully executable, day-by-day itinerary.

Rather than requiring users to piece together flights, lodging, activities, and transportation across multiple platforms, TripCraft provides a single, guided workflow. Users can start with a destination or with inspiration, customize priorities such as pace, budget, and comfort, and interactively refine their plans.

What makes TripCraft different is that it does not stop at generating ideas. The system validates each decision against real-world constraints—time, distance, and pacing—and produces itineraries that users can realistically follow. The final plan includes a trip overview, feasibility analysis, daily schedules with logistics, budget breakdowns, and direct booking links.

TripCraft is designed not just to inspire travel, but to make it actionable.

How we built it

TripCraft is built as a web-based application with a structured, multi-step planning flow designed around decision-making rather than conversation.

We initially experimented with the Google Gemini API by building a simple chatbot. While it could answer questions, we quickly realized that a purely conversational interface breaks down for complex planning tasks. Travel planning requires staged decisions, constraint propagation, and comparisons—not isolated answers.

To address this, we redesigned the system around a guided front-end experience. Early in the flow, the interface collects key travel constraints such as departure location, group size, travel dates, and travel pace. These constraints are passed directly into the Gemini API, so recommendations are generated with feasibility in mind from the start, rather than filtered afterward.

The system then guides users through destination discovery, transportation selection, lodging recommendations, and activity curation. Each step builds on the previous one, with the state maintained throughout the process.

In the final stage, TripCraft performs an explicit activity feasibility analysis. Instead of blindly scheduling everything, the system evaluates each activity against real-world constraints such as transit time, geographic distance, and daily pacing limits. Activities are categorized as scheduled, optional, or infeasible—and the system explains why certain activities cannot fit.

By combining a clean front-end, structured state management, and AI-driven reasoning, TripCraft moves beyond a chatbot and becomes an interactive planning assistant.

Challenges we ran into

One of the biggest challenges was preventing the AI from over-recommending. Early versions of TripCraft tried to fit every appealing activity into the itinerary, resulting in plans that looked impressive but were completely unrealistic.

Another major challenge was designing the feasibility analysis. Silently removing activities made the system feel arbitrary and untrustworthy. We needed the AI not only to make decisions, but to justify them in a way users could understand. Designing explanations that were accurate, concise, and human-readable required significant iteration.

Finally, integrating AI reasoning smoothly into the front-end flow required careful prompt design and state management. Each step needed to feel intentional and consistent, rather than random or disconnected—a nontrivial challenge when working with generative models.

Accomplishments that we're proud of

Designing a constraint-aware AI system that prioritizes feasibility over superficial completeness
Building an explainable feasibility analysis that clearly communicates why certain activities are excluded
Moving beyond a chatbot into a stateful, guided planning workflow
Creating a clean, responsive web interface without relying on heavy front-end frameworks
Delivering a realistic, executable itinerary that users can actually follow

What we learned

Through this project, we learned that building useful AI systems is less about generating impressive text and more about operating within real-world constraints.

We learned how important it is for AI systems to understand when to say no, and how to explain those decisions clearly. We also learned how tightly AI reasoning must be coupled with interface design in order to guide users toward confident decisions instead of overwhelming them with options.

Most importantly, this project reshaped how we think about AI products: the most valuable systems are not those that generate the most possibilities, but those that help people make better, calmer, and more informed decisions.

What's next for TripCraft

TripCraft is only the beginning.

In the future, we hope to expand the system with restaurant recommendations, collaborative group planning, and deeper budgeting tools. We are also interested in exploring how the feasibility framework can be extended to other decision-heavy domains beyond travel.

More broadly, TripCraft represents our vision for AI-assisted decision-making: systems that reduce cognitive load, respect real-world constraints, and help people enjoy the experiences that matter most.