A platform for speaking clubs where members book one-on-one sessions with each other to practice languages for free
Everyday practice is the only way to master a foreign language after you reach the A2 level. Teachers cost a lot, and students need speaking partners. I built a website where you can launch a club and let members book sessions through a shared calendar.
Solo Founder / Builder (2025):
I built the platform myself using Cursor + Supabase + OpenAI API + Replit + Firebase + Telegram bot + Brevo.
Repacked it as a B2B SaaS for language schools and career agencies.
Found B2B customers through cold outreach.
Launched the service and iterated with real customers.
Integrated an LLM API that doubled the platform’s audience.
Create your speaking club where members book 1n1 sessions
Why
Millions of people get stuck at the B1–B2 level (like I did) and hesitate to interact in a foreign language because they lack confidence and fluency.
Learning apps let you practice rules, not information exchange.
Private tutors let you practice for ~$40 per session — an amount you won’t pay every day.
There were no efficient services for finding everyday speaking partners you actually have something to talk about with.
Business
Monetised as a B2B SaaS for language schools and career agencies to engage and retain potential and former clients. $99/month for groups of up to 200 members.
MVP growth logic: focus on helping existing communities one by one because of the cold-start problem.
Found the first customers through cold outreach via targeted emails.
Product
Lets communities create closed group calendars where users book 1-on-1 sessions.
Authorisation via Google with ToU & PP acceptance.
Scheduling logic: sharing time slots, requesting, booking, and cancelling sessions with other members.
Notifications via email or Telegram bot.
The chat system lets users reschedule meetings without revealing personal contact details.
The profile page lets users set up languages, interests, a LinkedIn link, and notification preferences.
P2P mock interview mode lets users practice job interviews with peers and generate personalised questions using an LLM.
Share recurring time slots for long periods or add specific time preferences
Get notifications and chat via email or Telegram
To engage a new type of audience — people who practice mock interviews — I implemented personalised agenda creation. It lets users generate unique sets of questions they want to be asked based on their CV and job description. OpenAI API, this was your turn.
Practice p2p mock interviews with personal questions that are based on your CVs
How I built it
After creating a demo-ready MVP, I found a company ready to become my design partner and build the service together. They also became my first client.
Simplified process for any feature:
Dived into a new technical subject (such as authorisation tokens or cron jobs) with ChatGPT.
Described the detailed pipeline to Claude via Cursor. The level of detail required describing each query because LLMs weren’t good enough for consistent autonomous work in mid-2025.
Some of the most frequently used pipelines on the platform:
Most active members gathering. Yaroslav shares how he just found a job abroad