Launching today

Gadfly
The technical interview tutor that asks instead of telling
1 follower
The technical interview tutor that asks instead of telling
1 follower
Gadfly is a voice tutor for technical interview practice. You work through algorithms, system design and behavioral rounds out loud; the tutor answers with questions. It names what broke, remembers the weakness and brings it back until it clears. It quotes your words and the minute when it can, and says nothing if the record is too thin. For engineers facing big-tech interviews. 2 free sessions, no card. Laptop or desktop, Chrome or Edge. Built for preparation, never live-interview assistance.









Free Options
Launch Team

ClarifyThe full-stack AI CRM for modern GTM teams
Promoted
Hi Product Hunt, I'm Dmytro. I'm a software engineer and architect with 18 years in IT, and Gadfly is the first product I've built and shipped on my own. This is my first launch.
Why this idea: I was preparing for a big-tech loop myself. I had the usual list of solved problems, and I still froze the moment someone asked "why this approach?" out loud. A chat gave me the answer every time I got stuck, I nodded, and a week later I made the same slip again. What I needed was not another answer. I needed someone who asks, keeps a record of where I break, and brings it back until it stops breaking. Nobody was selling that, so I built it.
Gadfly is a voice tutor for the three rounds, and a session goes like this:
Pick the round, algorithms, system design or behavioral, and talk it through out loud (or type). The tutor answers with questions. "I'm stuck" gets you one more step, up to pseudocode; "tell me the answer" gets a no and a next step. In system design you commit to your design before the model answer unlocks.
At the end you score yourself first, then see the tutor's score and the gap. It names what broke and quotes your own words and the minute when the record has them. If the record is too thin, it says nothing rather than inventing a finding.
The slip goes into a journal the model doesn't write. It stays open until three clean runs close it, and the next problem is picked against that journal, with the reason shown. In behavioral rounds a decision counts only when it was yours: a story that never leaves "we" is not credited.
To be fair: ChatGPT or Claude can run a good mock tonight, and Gadfly's tutor runs on the same kind of model. The difference is the record between sessions. If you already keep that journal by hand after your mocks, you don't need Gadfly.
Good to know: 2 sessions free, no card. Then one-off packs, 8 sessions for £19.99 or 16 for £35.99, VAT included, no subscription, 14-day refund. Laptop or desktop, Chrome or Edge. The first session runs in text; the voice (about 1 GB) downloads once and runs in your browser tab, so raw audio never leaves it. Code in Java or Python. For practice before an interview, never during one.
The most useful thing you can give me is honesty: run one session on a problem you actually bombed once, then tell me where the diagnosis was wrong, or where the tutor gave away too much.