Launching on Product Hunt in 2 days — solo, from Japan, no code background. What do first-timers usually screw up?
Hi IH! Solo founder from Tokyo. I spent my career in logistics — no engineering background, no network, and honestly limited English (writing this with AI's help). Yet I've shipped 6 products since March by directing AI, mostly by speaking into a microphone.
My next one, Kotoba Dojo, goes live on Product Hunt this Tuesday. It's a game that trains how precisely you can describe things to AI: you describe a target image in your own words, AI draws only what you said, and you get scored on the match. The game itself was built the same way — by talking.
Coming Soon page: https://www.producthunt.com/products/kotoba-dojo
This is my first PH launch. For those who've been through it — what do first-timers usually get wrong on launch day? Anything you wish you'd known?
I'll report back with results either way 🥋
For me, the most interesting part is this: Kotoba Dojo is not just a game. It's a good way to practice one of the hardest skills with AI — describing exactly what you want. I teach a free course for complete beginners, mostly older people with no tech background, who want to learn ChatGPT, Claude, and Gemini. Most of them get stuck because their descriptions are too vague. Did you ever think about using the belt ladder as a teaching tool for total beginners, not just as a game for people who already know how to prompt
Really inspiring story, especially how you’ve built and shipped 6 products by directing AI. I’m also very interested in AI-first product building.
I’m currently looking for an opportunity to work with a startup/founder where I can contribute and learn. I’d love to show you MindPulse and Natour AI to give you a sense of what I can do.
If you’re open to it, I’d really appreciate the chance to connect and hear if there might be any opportunity to work together. Wishing you the best for the Product Hunt launch!
Thank you for the kind words, Kiray! I have to be upfront — I'm a solo founder running everything bootstrapped, so I'm not in a position to bring anyone on right now. My whole approach is directing AI agents instead of hiring, which is kind of the point of the story 😄
That said, best of luck with MindPulse and Natour AI — keep shipping!
Really inspiring story, especially how you’ve built and shipped 6 products by directing AI. I’m also very interested in AI-first product building.
I’m currently looking for an opportunity to work with a startup/founder where I can contribute and learn. I’d love to show you MindPulse and Natour AI to give you a sense of what I can do.
If you’re open to it, I’d really appreciate the chance to connect and hear if there might be any opportunity to work together. Wishing you the best for the Product Hunt launch!
Six products shipped by speaking into a microphone is the detail I would not bury, that is the story.
The thing first timers usually get wrong is not the copy, it is that launch traffic hits states nobody built. A game like Kotoba Dojo has three: what the screen shows while the AI is drawing, what a new player sees before their first round exists, and what happens when the image generation call fails or times out. Prompted builds tend to only have the happy path, so hour one of a launch is when strangers find the other two.
Worth walking your own product once and forcing each of those on purpose before Tuesday. My checklist for that pass is here: https://durablefoundations.gumroad.com/l/pyramid-reality-check
Which of those three states have you actually seen with your own eyes?
Kael Voss / DurableFoundations
Fair challenge. All three, actually — and two of them broke during testing, which is why they're handled now: there's a step-by-step loading state while the AI draws (with a visible progress path), a first-run tutorial + a full sample round in the gallery before you ever spend a credit, and failed/timed-out generations auto-refund the credit with a retry prompt. The refund logic is server-enforced, not client-side.
Launch traffic will surely find something I missed, but the unhappy paths were exactly where the belt-gating bugs showed up in week one, so they got the most testing time.
The strongest signal is that the founder has already shipped six products with essentially no-code execution. The interesting question now isn't whether he can launch—it’s whether Kotoba Dojo has a specific enough user problem and retention loop to turn a Product Hunt spike into sustained usage.
That's the right question. My retention bet is the belt ladder: 50 levels with a yokai "gatekeeper" boss at each belt, and the sensei's feedback explains which specific words changed the picture — so each round teaches something concrete instead of just scoring you. Whether that loop holds beyond the launch spike is exactly what I'm watching this week. Day-1 data will tell.
That’s a meaningful retention signal. I’d be interested to see what the first week tells you about whether the loop holds beyond the launch spike.
Deal — I'll post week-1 numbers in this thread. The two I'm watching: how many players come back for a second session, and how far past yellow belt the median player gets. If the loop only survives on launch-day curiosity, those two will expose it fast.
That’ll be interesting to watch. I’d be happy to continue the conversation as you see the numbers come in — would you be open to sharing the best email to reach you on?
Appreciate that! I'll keep posting the numbers right here in this thread so everyone can follow along — that's the fun of building in public.
For anything beyond the thread, easiest ways to reach me are X (@letizia_inc) or the contact form on the site. Looking forward to comparing notes as week 1 unfolds.
Makes sense — I’ll keep an eye on the week-one numbers here. Looking forward to seeing what the data says.