OpenDoor

Inspiration

OpenDoor was inspired by a simple but important problem: discovering a culturally interesting place does not mean that the place is physically accessible. For wheelchair users, people with mobility needs, and caregivers, missing or unreliable accessibility information can make planning an outing stressful and uncertain. I wanted to build a solution that connects cultural discovery with transparent, independently sourced accessibility evidence.

What It Does

OpenDoor is an accessibility-aware cultural discovery agent designed to combine Qloo's cultural taste intelligence with independent venue accessibility information. Its core principle is that cultural relevance must never be treated as proof of wheelchair accessibility. Instead, the system distinguishes verified, partially verified, unknown, and conflicting evidence so users can make more informed decisions.

How I Built It

I developed OpenDoor using Next.js, TypeScript, React, Tailwind CSS, and Vitest. Its seven-stage workflow validates requests, resolves cultural interests, connects to the Qloo API, normalizes venue information, matches accessibility evidence, ranks results transparently, and generates an inspectable audit trace.

I also designed a preview mode so the application can remain usable when API credentials are unavailable. The architecture keeps API credentials on the server and separates cultural recommendations from physical accessibility verification.

Challenges and Learnings

One of the main challenges was designing a system that does not make unsupported accessibility claims. Cultural recommendations, popularity, and similarity scores cannot establish whether a venue has step-free entry or wheelchair-accessible facilities. This required a conservative evidence-matching approach and clear communication when information is missing or conflicting.

Building OpenDoor strengthened my understanding of agentic workflows, API integration, transparent decision-making, and responsible AI design. It reinforced an important lesson: useful AI should not only make recommendations but also communicate the evidence and uncertainty behind them.

Current Status and Future Work

OpenDoor has been deployed for public testing, and its automated test suite has 13 passing tests. At the time of submission, I am still awaiting my event-issued Qloo API key, so live Qloo integration has not yet been verified. Once access is available, my next step is to test the live integration and validate the complete discovery workflow.

My goal is to make cultural exploration more inclusive by helping people discover places that match their interests while making accessibility evidence and uncertainty visible before they visit.

Built With

• api
• next.js-15
• qloo
• react
• tailwind-css
• typescript
• vitest