Skip to content

Quentin Bottos-Ecrepont

⁠All work

FrancoTokyo

Jobs for French speakers in Japan were scattered everywhere, so I put them in one place. FrancoTokyo now gets 9,100+ visitors a month with no paid ads, and employers pay to feature their listings.

The job platform for French speakers in Japan, built to be found on Google and by AI assistants.

Role
Founder
Context
Independent project
Period
- Present
Location
Tokyo

TL;DR

Finding a job at home is hard. Now try finding one in French, in Japan. The listings exist. They're just scattered everywhere.

So I built FrancoTokyo to put these jobs in one place. I decide what gets built and test it, and AI agents (Claude Code) write the code.

The hard part was getting found without paying for ads. So I designed the site to show up on Google and in AI assistants' answers.

Every month, it gets 9,100+ visitors and sends 2,250+ candidates to employers. And employers now pay to feature their listings.

BFM Business and Le Journal des Français à l'étranger interviewed me as its founder, and Anthropic selected it for its Claude for Startups program.

The problem

A job board has two sides, and each one waits for the other. Job seekers go where the jobs are. Employers post where the candidates are.

Someone has to make the first move.

So I started with the jobs: I find listings, check them and rewrite them by hand.

But finding a job is only part of the problem. You still need to apply, prepare your CV and understand how recruitment works in Japan.

So I added job alerts, guides about working in Japan and AI career tools.

On the other side, employers get a recruiter dashboard and a database of candidates.

My role

I define the product and design the UX and UI. I also handle SEO and GEO, and talk to employers.

I grew the community one post at a time. I posted in French-speaking Facebook groups and on LinkedIn, and teamed up with partners across the French-speaking professional community in Japan.

Early on, I wanted to query the data in SQL. So I decided to move the first prototype from Firebase to Supabase, which runs on Postgres. Every table has row-level security, so Supabase's public API can't read the database.

How people find it

  • Google

    • Pages for each category and city
    • Guides about working in Japan
    • Listings in Google for Jobs
  • AI assistants

    • AI-readable site guide
    • FAQs as structured data
    • MCP connector in Claude's directory
  • Community

    • Facebook groups and LinkedIn
    • Telegram bot and newsletter

FrancoTokyo, a Next.js and TypeScript app

  • For job seekers

    • Job listings, search and alerts
    • Credit-based AI career tools
    • Free visa application analyzer
    • Applications and saved jobs
  • For employers

    • Post a job
    • Recruiter dashboard
    • Candidate database
    • Featured listings
  • For the community

    • Housing listings
    • Events
    • News
  • Behind the scenes

    • Job listing aggregation
    • Checked and rewritten by hand
    • Moderation and analytics

What makes it work

  • Database

    • PostgreSQL and row-level security
  • AI inference

    • Career tools and document analysis
  • Payments

    • Featured listings and credits
  • Email

    • Notifications and job alerts
  • Authentication

    • Accounts and permissions
  • Security

    • CDN, bot protection and rate limits
  • Infrastructure

    • Hosting and automated deployment
  • Analytics

    • Traffic and listing stats
Drawn from FrancoTokyo's codebase.

How I work with the agents

I describe what I want, and an AI agent (Claude Code) builds it. I test the result, point out what's wrong, and we go again.

Key decisions

Build pages people are already looking for

I wanted FrancoTokyo to be found without having to pay every time someone clicked on a link. So I bet on organic search. The site generates a page for each job category, and one for each category in each city: tech jobs in Tokyo, finance jobs in Osaka. Each with its own title and description.

the long game

Together with the guides and the press coverage, these pages bring in the search traffic.

A connector for AI assistants

AI assistants already send 620+ visits a month to FrancoTokyo.

So I added a connector built on MCP, the open standard that lets an AI assistant use another service. Ask Claude for jobs in Tokyo that don't require Japanese, and it searches FrancoTokyo's listings and links to each one.

Every answer links back to the listing on FrancoTokyo, so visits and applications still count. It went live in Claude's connector directory the day I submitted it.

A free Working Holiday visa analyzer

Member profiles showed me many people on a Working Holiday visa. Getting that visa starts with an application file.

So I added a free analyzer for Working Holiday applications. Upload the PDF with your cover letter and stay plan, and get an analysis back.

Let the data tell me what needs fixing

The data showed me where new members dropped out: right before the screen where they pick a profile.

So I fixed the sign-up flow, one step at a time. Members who get through that screen went 1 in 3 → 9 in 10. Those who upload a CV went 1 in 7 → 1 in 3.

1. Sign in

  • Google or an email link

    • No password to create

2. Choose a profile: job seeker or recruiter

  • Before

    • Sign-in sent new members to the home page
    • Most never saw this screen
  • The fix

    • Every sign-in link leads to this screen

3. Fill in the profile

  • Before

    • Job seekers landed on a dashboard
    • A profile gets filled in the first minutes, or never
  • The fix

    • The profile form opens right away

4. Upload a CV

  • Before

    • A CV alone didn't make the profile visible to recruiters
  • The fix

    • The profile fills itself from the CV
    • Visible to recruiters in one click
    • Free credits for the first PDF CV
Each step and its fix, from FrancoTokyo's product decision log.

Career tools built for Japan

Applying for a job in Japan has its own codes: a Japanese résumé, a detailed career history, and recruiters who check your Japanese level and visa status first.

So I added 8 AI career tools specifically for the Japanese job market, paid with credits.

AI career tools for job seekers

  • CV analysis

    • Read as a Tokyo recruiter would: score, checks and action plan
  • ATS simulator

    • Formatting, sections and missing keywords
  • Job match

    • Your CV against a listing, with the gaps
  • Cover letter

    • Written for Japanese business culture
  • Interview prep

    • Questions from your CV and the listing, with feedback
  • LinkedIn optimizer

    • Profile improvements for the Japanese job market
  • Rirekisho

    • The Japanese résumé, in the standard format
  • Shokumu keirekisho

    • A detailed career history, in French and Japanese
Every tool is live on FrancoTokyo.

Upload your CV as a PDF, and you get free credits to try a tool. The first job seeker to use them ran the ATS simulator, then bought credits 4 minutes later to run the CV analysis.

Not everything needs AI

Every CV analysis ends with a score. Ask an AI to score the same CV twice and you can get two scores. A rule gives the same score every time, and can say why. So I moved part of the scoring from the AI to fixed rules.

boring on purpose

The AI still rates each check: Japanese level, right to work in Japan, presence in Japan, career continuity and readability. Those ratings held from one run to the next. Only the number moved. Now the code turns them into the score, with fixed points for each check.

Before, the same CV scored anywhere from 68 to 80. With the rules, it got the same score in 12 out of 12 analyses.

Japan market fit

Japan market fit: 63 out of 100

A Tokyo recruiter's checks

What a recruiter looks at before reading your CV in detail.

  • Japanese level18 of 30 points

    Most jobs at Japanese companies ask for N2 or above.

  • Right to work15 of 30 points

    A valid visa, or the degree or experience an employer needs to sponsor one.

  • Presence in Japan10 of 20 points

    Already living in Japan, or a planned arrival date.

  • Career continuity10 of 10 points

    Gaps, unexplained overlaps, several jobs shorter than a year.

  • Readability10 of 10 points

    Full dates, clear job titles, typos, mixed languages.

A sample CV analysis, translated from French. Change a rating to see how the score moves.

Results

As of :

9,100+
visitors a month
2,250+
candidates sent to employers every month, through "Apply" or directly on FrancoTokyo
First paying employer within a month of launch, with no outreach

Featured listings are self-serve: an employer pays through Stripe, without contacting me.

Traffic comes from search engines and AI assistants:

65,000+
page views a month, with no paid ads
3 in 4
referred visitors come from Google
620+
visits a month sent by AI assistants, mostly ChatGPT, ×2.8 in one month

On the job seeker side:

710+
registered members
260+
job alert subscribers
15%
of job views turn into a candidate sent to the employer

On the employer side:

450+
job listings, checked and rewritten by hand or posted by employers
275+
companies
150
CVs available to employers, ×5 since early August
19
applications with CVs in 9 days on a tour operator's featured listing

Press and recognition:

Interviewed on BFM Business TV

FrancoTokyo on BFM Business

BFM Business, Good Morning Business. Report by Astrée Olivier.

Interviewed in Le Journal des Français à l'étranger
Source (opens in a new tab)⁠
Selected for Anthropic's Claude for Startups program

Next steps

FrancoTokyo started with jobs, but moving to Japan involves a lot more than finding work.

Ubigi, an eSIM provider, signed as a partner. Other companies that serve foreigners in Japan, housing providers among them, have agreed to work with FrancoTokyo, and we're setting up those partnerships now.

Next, I want more employers to post their jobs directly.

When it bought a featured listing, an international tour operator also asked for my placement terms. In Japan, charging placement fees takes a license from the Ministry of Health, Labour and Welfare. With it, placement would become a second business model.