CiaoTutor
Finally speak the Italian you already understand.
I hit the intermediate plateau and got stuck there for a long time. Duolingo, language schools, and friends got me through the basics. I could read, I could pass grammar quizzes, and then I'd sit across from an actual Italian person and freeze. Or, best case scenario, talk like a 5-year-old kid. The gap between "understands Italian" and "speaks Italian" is enormous, and almost nothing is built for the B1–C1 range where that gap lives.
So I built CiaoTutor (the tutor's name is Elena). The headline feature is that you talk to her, a real voice conversation, no tapping words into the right order. But talking alone doesn't teach you anything; I learned that the hard way. The actual learning is in the loop around the conversation:
She corrects you silently, mid-conversation. Instead of interrupting you every time you fumble a subjunctive, mistakes get logged in the background so the conversation keeps flowing. You stay in the deep end.
Every session ends in a report. Your errors are grouped by type, new vocabulary, and a CEFR-anchored listening score, so you can actually see what you got wrong and what you're improving on.
It generates exercises from your mistakes. Not generic drills, targeted reading, writing, and speaking exercises built from the specific things you got wrong that session.
It remembers you. Recurring mistakes and past sessions carry forward into the next conversation, so Elena adapts instead of resetting to zero every time.
She has a personality. You're talking to someone, not a neutral robot — which turns out to matter a lot for actually wanting to come back tomorrow.
The conversation is the fun part; the reports, exercises, and memory are what make it stick. That whole loop is the thing I could never get from an app before.
It's live at ciaotutor.com and still early. I'd genuinely love feedback from anyone who's been stuck at that same "I understand it but I can't produce it" wall. Does this loop match how you'd want to break through, or am I missing something?
Happy to go deep on the build too (it's a real-time voice bridge to a live speech model, which was a fun/painful engineering problem).
5 Comments
What stood out to me is that you designed the learning around the moments people get stuck, not around the conversation itself. I'd keep validating whether users come back because they enjoy talking to Elena or because they can clearly see themselves becoming more confident speakers.
Hi Aryan, it seems to be a bit of both from the feedback I've received so far. The session's progress report and exercises really push users to learn and keep evolving, but some of them confessed they usually skip the exercises and just like to talk to Elena and follow their progress through the reports.
Thanks! I’ve just sent it over.
Looking forward to hearing your thoughts whenever you have a chance.