Landed

Inspiration

Every year, thousands of students leave everything they know to study abroad on scholarships — and land alone, overwhelmed, googling everything from scratch at midnight in a country where they don't even know how to buy a SIM card. We built Landed because we wanted to be the friend who already lives there — the one who tells you exactly what to do, in your own language, the moment you land.

What it does

Landed is an AI companion for first-time scholarship students moving to the UK, Canada, or Germany. After a 30-second setup (name, country, language), students get a personalized dashboard with six guided topics — First 72 Hours, Housing, SIM & Bank, Transport, Budget, and Documents — each grounded in real, verified research. They can tap a suggested question or type their own, and the AI responds conversationally in their chosen language, always reminding them to confirm anything binding with their university.

How we built it

We built Landed as a single-file React app (no build tools, no bundler) to keep it lightweight and reliable on low-bandwidth connections — a real constraint for us, building entirely on a phone over 3G in Lagos. The AI companion runs on Google's Gemini 2.5 Flash API (free tier), grounded by research our team manually compiled and verified across UK, Canada, and Germany. Claude (Anthropic) assisted us throughout — architecting the app, writing the React code, and helping us debug issues live.

Challenges we ran into

Our biggest challenge wasn't the AI — it was infrastructure. We tried building on Bolt.new first, but our 3G connection couldn't reliably push deploys through, and we lost hours to stuck builds and blank screens. We pivoted to hand-writing the app directly in plain React with no build step, which turned out to be far more reliable. We also hit a wall when our first AI provider required paid credits we didn't have — we switched to Gemini's free tier, debugged a wrong model name causing a zero-quota error, and built an automatic fallback so the app always works even if live AI calls fail.

Accomplishments we're proud of

We built a fully working, multilingual AI companion — entirely on a mobile phone, over 3G, with zero budget — that gives specific, verified, country-grounded answers instead of generic advice. We're especially proud of the responsible AI guardrail baked directly into the system prompt: every single response reminds students to confirm binding decisions with their university, not as a disclaimer, but as a real design choice.

What we learned

We learned that resilience matters more than perfect tools — every time one platform failed us, we found another way through. We also learned how much specific, grounded information matters for trust: a generic AI answer about "renting abroad" means nothing to someone who needs to know exactly what a UK tenancy deposit cap is, today.

What's next for Landed

Next, we'd build deep API integrations directly with local digital banks and verified student housing providers so students can begin setting up their lives before they even board the plane. We're also planning a peer-to-peer community hub connecting incoming students with seniors who've already been in their shoes.

Built With

• claude
• css3
• google-gemini-2.5-flash-api
• html5
• javascript
• netlify
• react