Build Story · Relocation atlas · Web
Awaye: an atlas with an opinion
A relocation finder that scores 12,000+ cities against your budget and passport, and puts each match's weakness in writing.
In one look
The problem
People who have spent years googling "where should I live" keep finding the same ranked lists of cities. None of them knows your budget or your passport, and none says what could go wrong.
What we built
Say where you're tired of being in one sentence for a rough cut, or take a four-part quiz. Awaye scores 12,000+ cities, uses your budget and passport as filters, and gives every match a dossier and a weakness card.
Where it is now
Live and free at awaye.app, with no account needed to search: matches, city pages, head-to-head comparisons, a couples mode and an open-formula ranking of 15 nomad cities. Reaching the people it was built for is the work now.
What this would have cost in 2024
About $35k and 2–3 months
2024 team: 3 people: full-stack engineer, data engineer, product designer with QA. Built here by one founder working with AI in 11 days to launch, 36 working days in all.
| Piece | 2024 hours |
|---|---|
| Matching engine: semantic prescoring, budget and passport filters, diversity-aware top 5, feedback penalties, tests | ~207 |
| City data pipeline: 12,000+ city catalogue, scheduled AI enrichment, visa rules, neighbourhoods, climate | ~177 |
| Quiz, results and dossier UI: globe, weakness card, sharing, couples mode | ~187 |
| Public pages: city pages, head-to-head comparisons, why-not pages, nomad ranking, sitemap, share images | ~93 |
| Accounts, saved searches, public profiles, weekly email digest | ~73 |
| Admin, monitoring, rate limits and security passes | ~73 |
| Brand and landing: name, wordmark, visual system, landing page, postcards | ~73 |
| QA, smoke tests, caching and search clean-up | ~42 |
| Total (PERT) | ~926 |
Method: PERT (O+4M+P)/6 per piece; low = O × $25, likely = PERT × $37.5, high = P × $50 (Accelerance 2024 Eastern-Europe band). These are estimates of what a team would have needed in 2024, not invoices.
How it was built
From first commit to current build in 11 days to launch, 36 working days in all
- First commit
The idea: describe the life you want and get matched cities on a globe.
- A matcher in a day
Semantic prescoring, visa rules matched to your passport, a plain-numbers reality panel, save and compare, and retrieval tests.
225 commits that day - Filling in the cities
A scheduled job wrote city narratives in batches, neighbourhoods were filled in, and a live-signals panel was added to each place.
narratives for 3,787 cities in one run · 916 neighbourhoods backfilled - The quiz and the deal-breakers
A step-by-step quiz, deal-breakers as hard filters, a tap-water safety filter, confidence labels on results and a share image per match.
- Variety, weaknesses and couples
The top five became diversity-aware, the weakness card got its own place on each result, and couples mode runs the matcher for each partner and for both together.
- Named Awaye
The product got its name and a wordmark that tells you how to say it: "(with the e)".
- A ranking clean-up and comparison pages
Budget scoring, keyword search and the wildcard pick were fixed, the climate bonus was capped, and head-to-head comparison pages went live.
275 commits that day · climate bonus capped at ±10 - A landing page with an argument
The quiz moved to its own page. The new front page explains the problem, prices an ordinary Tuesday and says which of the ten gauges are estimates.
- Security pass
Paid AI and maps calls got rate limits and spending caps, and the framework was upgraded.
- City pages that admit what they don't know
Blocks with no data now hide instead of showing an error, community chatter comes from a new source, and thin pages were taken out of search.
community pulse cached: one paid fetch per city per 7 days
Who did what
The founder decided
- Every match ships its weakness, in writing.
- Budget and passport work as filters, not as soft preferences.
- Modelled scores are labelled as estimates, and anything without a source prints as unknown.
- An account is needed to save a search, not to run one.
- Thin pages stay out of search until they carry real content.
AI built and checked
- The web app, built in Lovable from the founder's briefs: matcher, quiz, city pages and the data pipeline.
- City narratives, neighbourhood data and weakness cards, written by AI from open data and labelled as such on the site.
- Claude Code: the security pass, the city-page fixes, the live-signal cache and the search clean-up.
Stack
Want something like this for your team?
From one call to something live. The same method is in the Claude Operator Pack.
From the founder
You might also like
Shares custom-domain
Shares custom-domain
Shares custom-domain