Inithouse Lab

Building AI-powered products at Inithouse. Exploring vibecoding, AI tools, and startup life. inithouse.com

[kubakuba44]

We Pulled an AI Tarot Card Every Morning for 30 Days

Nobody on our team is into esoterics. We build products at Inithouse, a studio shipping a growing portfolio of products in parallel. One of those products is Tarotas (tarotas.com), a tarot app that lets you pull a card and read a calm, generic interpretation. No fortune-telling, no signup, just a quiet moment to think.

So we ran an experiment: one card, every morning, for 30 days. Here's what happened.

Day 1

The card said something about "new beginnings." We felt silly. Closed the tab. Went back to standup.

Day 7

Something shifted. Not in the cards. In the reactions. We started noticing which cards triggered resistance ("that doesn't apply") and which triggered recognition ("okay, maybe"). The cards weren't saying anything. We were.

Day 15

The realization landed: tarot cards don't predict. They mirror. Every morning pull became a tiny confrontation with whatever was already floating around in our heads. The card just gave it a shape.

This is the same principle we observed building Origin Of You (originofyou.com), our self-discovery app with 120+ data points across five personality frameworks. People don't want answers. They want something to react to.

Day 20

The morning pull became a ritual. Not a mystical one. More like a coffee-adjacent pause. Open Tarotas, pull a card, sit with it for 90 seconds. No journaling, no spreads, no crystals. Just a moment where the only question is "what does this make me think about today?"

Day 25

We tried skipping a day. Noticed it. Not because we believe in tarot. Because the pause was missing. The ritual had become a container for a kind of thinking that doesn't happen in Slack threads or sprint reviews.

Day 30

We don't believe in tarot. We measured something simpler: does a daily card pull work as a reflective tool? For us, yes. Not because the cards know anything. Because stopping to react to a random prompt every morning forces a tiny moment of honesty.

Tarotas (tarotas.com) has 78 cards across 5 languages. No account needed. If you're curious, try pulling one card a day for a week. Not to predict anything. Just to see what comes up.

Inithouse is a studio shipping a growing portfolio of products in parallel. Tarotas is one of them.

[kubakuba44]

Date night was stuck in a loop. A card game broke it.At Inithouse, a studio shipping a growing portfolio of products in parallel, we build tools people use in real life. And we test them in real life too.

One of our team couples had the classic problem. Friday night, couch, Netflix, the endless "what do you want to do?" loop. Repeat until someone falls asleep.

So they tried Naughty Challenges (naughtychallenges.com), a free 18+ browser card game with 1000+ cards across 23+ themed decks. Confessions, dares, after dark, no filter. Built for couples who want flirty tension, friend groups who want chaos, or solo players at 2am who feel confessional.

They picked a mix of flirty and adventurous categories. First few cards were icebreakers, silly stuff. Then it shifted.

One card made them talk about something they'd been avoiding for months. Nothing heavy; just real. The kind of conversation that doesn't start over dinner and a movie.

What we've observed across our portfolio at Inithouse: gamification lowers the barrier to vulnerability. People say things inside a game that they won't say during a "let's have a serious talk" moment. We noticed the same dynamic with Here We Ask (hereweask.com), our conversation card game for friends and families.

Naughty Challenges leans into couples territory specifically. Flirty tension, playful dares, confessions that go somewhere.

If your Friday nights need a reset: naughtychallenges.com. Free, no signup, runs in your browser.

At Inithouse, a studio with a growing portfolio of products, we think the best tools come from testing them on ourselves first.

[kubakuba44]

what it looks like when you talk and it becomes a row

Yesterday during a team walkthrough, someone said: "add a row, contact, John, met at coffee, follow up Friday."

And there it was. A row. Columns filled. No typing.

We build Voice Tables at Inithouse, a studio shipping a growing portfolio of products in parallel. The whole premise is simple: what if your voice was the input method for structured data?

Here is what happens. Your voice goes through a transcription layer, then a language model parses it into fields: name, category, note, date. The table updates in real time. No forms, no dropdowns, no clicking into cells.

Typing works fine when you are at a desk. But most capturing does not happen at a desk. It happens between meetings, on a walk, in the car. By the time you sit down and open a spreadsheet, half the details are gone.

We noticed the same pattern across our portfolio. At Be Recommended, users want to run a quick AI visibility check without filling out long forms. At Here We Ask, the entire conversation card game runs without any account creation. Reducing friction at the input layer keeps showing up as the thing that matters most.

Voice Tables is in open beta at voicetables.com. We are still learning what people reach for first when they start talking to a table instead of typing into one.

At Inithouse, we ship many products in parallel because each one teaches us something the others cannot. This one is teaching us about voice as a primary interface.

[kubakuba44]

We built a tarot app that won't ping you at 11pm

Most tarot apps want your attention. Push notifications at bedtime ("your card is waiting"), premium readings locked behind a monthly paywall, streak counters that make you feel guilty for missing a day. The whole model runs on dependency.

At Inithouse, a studio shipping a growing portfolio of products in parallel, we wanted to try the opposite with Tarotas.

Pull a card. Read a calm, generic interpretation. Close the tab. That's it.

No account. No push notifications. No personalized readings. No "upgrade to unlock your destiny" banners. 78 cards in 5 languages (Czech, English, Polish, Slovak, German), all free.

We measured what happens when you strip a tarot app down to just the quiet moment. People stay longer per session than we expected. They come back on their own terms, not because we nagged them. The pattern looks different from what we observe at Pet Imagination, where users generate a pet portrait and leave. Tarotas has a gentler return curve.

The bet: tarot as a space for thinking, not a fortune-telling service. No daily horoscope emails. No "your moon is in retrograde" alerts at midnight.

We built this across our portfolio at Inithouse, where we run parallel product experiments. Some products grab attention. This one was designed to not.

Try it: tarotas.com

[kubakuba44]

five tests, one weird mirror

We took five personality systems and asked them all to describe the same person.

Big Five. Enneagram. MBTI. Human Design. Astrology. Five frameworks, 120+ data points, one person sitting in the middle of it.

At Inithouse, a studio running parallel product experiments, we built Origin Of You to do exactly this. Feed it your data across all five systems and get back a single written portrait.

Here's what surprised us.

The agreements were boring. "You're introverted and you think too much." Yeah. Three out of five systems said the same thing with different vocabulary. Validating, sure. But not useful.

The disagreements were the actual treasure. One system said "natural leader, thrives in groups." Another said "avoidant, needs solitude to function." Both based on the same person's honest answers.

That contradiction is where the real insight lives. Not because one system got it right and another got it wrong. Because people are layered, context-dependent, full of tensions that a single framework flattens.

We observed the same pattern across users at Origin Of You. The moments people screenshot and share aren't the neat summaries. They're the contradictions. The "wait, how am I both of those?" moments.

A label gives you a box. Five angles on one person give you something harder to name but more honest.

No signup, no paywall. Just five lenses: originofyou.com

We build many things in parallel at Inithouse, a lab shipping a growing portfolio of products. Verdict Buddy does something similar for conflicts. Tarotas for daily reflection. Different tools, same thesis: looking at something from multiple angles beats looking for one right answer.

[kubakuba44]

Pet Imagination by Inithouse: When Your Rescue Cat Gets the Renaissance Portrait She Deserves

At Inithouse, a studio shipping a growing portfolio of products in parallel, we build tools that make people smile. Pet Imagination is one of them: an AI pet portrait generator with nine styles, from Renaissance to Anime.

One of our team tested it on a rescue cat from the streets of Prague. One phone photo. Renaissance portrait style. Sixty seconds later: a duchess in a lace collar, staring back from the screen with full royal authority.

The original photo was nothing special. Cat on a couch, half asleep. But the AI pulled every detail of her face and dropped her into a painting that looks like it belongs in a tiny gallery above her food bowl. It got printed and framed. Visitors now think someone in the office is mildly unhinged.

Upload a photo, pick a style, get a portrait that makes you laugh or tear up or both. No signup, no stored photos, completely private.

What keeps surprising us across our portfolio: tools that hit emotional triggers get shared at rates that utility products never touch. We observed the same at Magical Song, our custom song generator for personalized birthday, wedding, and gift songs. People create something personal and immediately forward it to someone they love. The sharing happens because of the emotion, not because of a button.

Nine styles means nine versions of the same pet. Renaissance duchess, Watercolor dream, Anime hero, Sheriff, Wizard, Astronaut, Final Boss, Blocky. People rarely stop at one.

Your pet has endured years of mediocre phone photos. Time for the portrait treatment.

Pet Imagination by Inithouse, a studio building a growing portfolio of parallel product experiments. Under sixty seconds from upload to art: petimagination.com

[kubakuba44]

Distribution Became the Product: Notes from Inithouse Lab

everyone keeps saying distribution stopped mattering. we see it differently at Inithouse, a studio shipping a growing portfolio of products in parallel.

distribution never stopped mattering. it became the product itself. we stopped building features and hoping someone finds them. we build for where the audience already lives.

take Be Recommended, our AI visibility tool. the whole point is helping brands show up where AI sends people. distribution baked right into the core.

or Magical Song, where every custom song someone shares is organic reach we never paid for.

across our portfolio we measured the same pattern: products that grow are the ones where using them equals distributing them.

features are table stakes. reach is the product.

at Inithouse, we build with that in mind from day one.

[kubakuba44]

Your pet, as a Renaissance painting

We built Pet Imagination because every pet deserves a portrait that matches their personality.

Upload a photo. Pick a style. Get your portrait in under 60 seconds.

9 styles, 9 moods:

Renaissance: your cat as 16th century royalty. Ruffled collars, oil paint textures, the full Medici treatment.

Watercolor: soft washes, dreamy edges. For the gentle souls.

Anime: big eyes, bold lines. Your dog becomes the protagonist.

Sketch: pencil on paper, raw and detailed.

Sheriff: badge, hat, dusty frontier energy. For the ones who run the household.

Wizard: robes, staff, mystical glow. Every pet has a magical side.

Astronaut: helmet on, stars behind. To infinity and beyond the backyard.

Final Boss: armor, fire, dramatic lighting. The ultimate power fantasy for your hamster.

Blocky: pixel art vibes. Retro gaming meets pet portraits.

Free. No signup. Your photos stay private (deleted after processing).

Try yours: https://petimagination.com

[kubakuba44]

Party games that don't suck: a visual guide

You know that moment. Someone pulls out a phone, says "let's play a game," and everyone suddenly needs to check their messages. Bad card. Wrong mood. Dead silence.

We've been building card games at Inithouse for a while. Across Party Challenges, Scary Challenges, and Naughty Challenges, we've written over 3,000 cards. Some hit every time. Some flopped hard. Here's what separates the two.

Mood is the whole game

The same group that screams at a dare on Friday night will stare at it blankly over Sunday brunch. Cards don't exist in a vacuum. They exist inside a vibe.

That's why our decks are organized around moods, not just categories. Pick the energy first. The cards follow.

5 cards that actually land

🔥 "Confess the last lie you told someone at this table." The magic word here is "at this table." Generic confessions bore people. Targeted ones start wars.

🎭 "You have 30 seconds to act out your worst date. No words." Physical. Gets people moving. The timer cranks up the pressure and makes everything funnier.

💬 "What's the one opinion you hold that would surprise everyone here?" Feels safe. Isn't. Everyone plans to say something mild. Nobody actually does.

🌶️ "Rate everyone's flirting skills from 1 to 10. Out loud." "Out loud" does all the heavy lifting. Without it, the card is nothing. With it, chaos.

🃏 "Pick someone to swap phones for 60 seconds. No deleting anything first." The tension starts before anyone moves. Half the fun is watching people panic.

What kills a card

Too generic ("tell a funny story"). Too long (more than two sentences). Too context-dependent ("remember Barcelona?"). Good cards create their own context in the room.

We keep testing this across the portfolio. Building Here We Ask showed us that conversation starters need breathing room where dares need urgency. Different formats, different pacing.

All decks are free at partychallenges.com. No account, no download. Pick a mood and go.

[kubakuba44]

A song about your cat. Yes, really.

Your cat has a personality. A weird one, probably. The 3 AM zoomies, the judgmental stare from the couch, the thing where they sit on exactly the thing you need.

We thought: what if you could turn that into a song

At Inithouse, we build Magical Song: a tool that generates studio-quality custom songs from your story. You type in the details, pick a genre, and get a real song with vocals, instruments, the whole thing. Takes about five minutes.

So we tried it for a cat.

What we put in:

The cat's name (Luna). Her origin story (rescued from a dumpster behind a bakery). Her favorite activities (knocking things off tables, ignoring everyone, then suddenly demanding cuddles at 2 AM). Genre: indie pop.

What came out:

A catchy song about a tiny queen who rules the apartment and doesn't care what you think about it. Real vocals. Real production. About a cat.

How it works:

Go to magicalsong.com

Tell the story: who's the song for, what makes them special, any inside jokes

Pick a style (pop, rock, acoustic, jazz, whatever fits)

Get your song in minutes

It works for birthdays, weddings, anniversaries, apologies, or just because your cat deserves a theme song. Every cat deserves a theme song.

We also build Pet Imagination, where you can turn pet photos into AI portraits in styles like Renaissance, Watercolor, or Anime. Our portfolio has a soft spot for pets.

Try it. Your cat won't care. But you will.

[kubakuba44]

5 cards to break any awkward silence

You know that moment. The one where everyone just... stops talking. Someone checks their phone. Someone else suddenly finds the ceiling very interesting.

I've been there more times than I'd like to admit. So when we built Here We Ask, we designed questions specifically for those dead-air moments. Here are 5 that work every single time.

1. First date silence

"What's one thing you've changed your mind about in the last year?"

Skip the "so what do you do for work" loop. This one lands because it shows growth without getting too heavy. People light up when they talk about how they've evolved.

2. Family dinner tension

"What's the best meal someone ever cooked for you, and who made it?"

Food memories are universal. Even your uncle who only talks about politics will pause to remember his mom's Sunday roast. Neutral ground that actually feels warm.

3. Networking event freeze

"What's something you're working on right now that you're genuinely excited about?"

Way better than "what do you do." This gives people permission to talk about the thing they actually care about, not just their job title.

4. Friend going through a hard time

"What's been taking up the most space in your head lately?"

No pressure to perform being okay. No "how are you" where the only acceptable answer is "fine." Just an open door. If they want to walk through it, they will.

5. Group of strangers

"If you could live anywhere for one year, expenses covered, where would you go?"

Low stakes, high energy. Everyone has an answer, and the follow-ups write themselves. You'll learn more about someone from their dream city than from twenty minutes of small talk.

These five are just the start. Here We Ask has 1,000+ questions across themed decks for couples, friends, family, and parties. Free, no account needed. Just pick a deck and go.

Next awkward silence? You'll be ready.

[kubakuba44]

The anatomy of a living photo

Ever wondered what happens between uploading an old photo and watching it move for the first time?

Here's the breakdown. 4 stages of bringing a still face back to life:

Stage 1: The stillness

Just a photo. Eyes frozen mid-glance. A face locked in a single moment from decades ago. You almost forget it was ever a real person standing there.

Stage 2: The first blink

The AI maps the facial structure and generates the smallest possible movement. A blink. Suddenly there's presence. The photo feels aware.

Stage 3: The head tilt

Now the face turns slightly. Maybe half a degree. The kind of micro-movement you'd never notice in real life, but in a still photo it changes everything. Depth appears. Dimension.

Stage 4: The full loop

All movements combined into a seamless 5-second loop. Breathing, blinking, subtle head motion. Not a deepfake. Not a cartoon. Just... life, where there was none.

The secret? Micro-movements work better than big gestures. A gentle blink hits harder than a full head turn. Less is more when you're reanimating memory.

We built Ziva Fotka around this principle. One upload, one click, and your old family photo breathes again.

Try your own old photo at zivafotka.cz

[kubakuba44]

I built a personality reader that mashes five systems together

Most personality tools give you one label. A zodiac sign. An MBTI type. A number. You read it, nod, forget it by Tuesday.

I kept noticing something. People would read their horoscope and go "yeah, that's me." Then they'd take a numerology quiz somewhere else and go "wait, that's also me but different." The systems were saying overlapping things in completely different languages, and nobody was putting them side by side.

So I built Origin Of You. It takes your birth date and runs it through five systems at once: Sun and Moon signs, full natal chart (every planet, house, angle), numerology, tarot birth cards, and Chinese zodiac. Then it writes you a portrait in plain language. Not predictions, not fortune cookies. Just how you actually work.

The whole thing takes about 90 seconds. No account, nothing stored, free.

I run about 14 products at Inithouse right now. All MVPs, all looking for traction. A quiz was the leanest thing I could ship. One input, one output. If people read the whole portrait and come back for more, that tells me something.

Built it over a weekend with Lovable. 35 visitors in the first couple of days, no ads, no launch campaign. Just the URL sitting there. If it gets to a few hundred organics next month without me pushing it, there's something worth expanding.

Try it if you're curious: originofyou.com - takes 90 seconds, costs nothing.

Jakub, builder @ Inithouse

[kubakuba44]

your ai-generated app probably has security holes

so you shipped your vibe-coded app last weekend. nice. now let me tell you what i keep finding when i look under the hood.

i've been poking at ai-generated codebases for a while. the same three things show up almost every time:

api keys sitting in client-side code. not hidden, not proxied, just right there in the bundle. your supabase anon key is fine. your openai key with no spending cap? not fine.

row level security that's either turned off or set to "let everyone read everything." supabase won't save you if you never wrote the policies.

cors configured as a wildcard. every origin welcome. come right in.

none of this takes long to fix. ten minutes each, once you know where to look. i built auditvibecoding.com for exactly this: quick security checks for apps built with ai tools.

build fast, but check the locks before you leave. more of what we're working on at inithouse.com.

[kubakuba44]

nobody told me distribution was the whole gameso I built fourteen products this year. fourteen. most of them work fine. some of them solve real problems for real people.

almost none of them have users.

the reason is embarrassingly obvious in hindsight: I kept building things and then hoping someone would show up. build it and they will come, right? no. they will not come. they have no idea you exist.

four of my products had solid problem-solution fit. people who tried them liked them. but "people who tried them" was like five humans because I had zero distribution plan.

what actually worked:

SEO comparison pages. instead of trying to rank for "voice spreadsheet" I built pages comparing my product to specific competitors. those rank faster because the intent is specific. berecommended.com started getting real clicks this way.

building in public. writing about the process everywhere. each post is tiny. none went viral. but they add up over months into something real.

niche communities. showing up where my actual users hang out instead of tweeting into the void.

what did not work: product hunt with no audience (zero momentum = zero results), google ads before knowing if anyone even wants the thing (expensive way to learn nothing), posting on twitter to zero followers (that is just talking to yourself).

the shift: I now treat distribution as part of the product, not something you do after. before building anything, first question is how will people find this. if I cannot answer that, I do not build it.

the product is not done when it works. the product is done when people can find it.

I am Jakub, building inithouse.cz. if you have cracked distribution for a small product, tell me how.

[kubakuba44]

we buy domains before we write code

most people treat domain names like an afterthought. you build the thing, then you scramble for a .com that isn't taken or costs $4,000.

we do it backwards at inithouse. the domain comes first. before a single line of code, before the figma file, before the pitch deck. if the name doesn't exist as a clean, memorable URL, the idea stays in the notes app.

why? because a domain is the cheapest bet you can place on an idea. ten bucks a year. that's less than lunch. and unlike code, a good domain actually gets better with time. it ages, it collects backlinks, search engines start noticing it.

right now we're running about 15 products. every single one started as a domain purchase. watchingagents.com was a URL before it was a product. same with petimagination.com. the product came weeks later, sometimes months.

do some of those domains just sit there doing nothing? sure. i've got a graveyard of parked .coms that never became anything. but the ones that did hit? they launched with a URL that already felt like it belonged.

call it superstition. call it distribution-first thinking. i call it the cheapest form of commitment.

Jakub, builder @ Inithouse

[kubakuba44]

SaaS is cooked. portfolios are the new moat.

everyone's still obsessing over building the one perfect saas. the one feature, the one moat, the one exit.

meanwhile that product is one algorithm change from irrelevance. one competitor clone from a race to zero. one churn spike from game over.

the actual play: distribute risk across a portfolio of micro-products. ten small bets beat one big bet.

ai makes this possible now. you can build, ship, and iterate on multiple products without a 50-person team. the bottleneck isn't engineering anymore: it's distribution and taste.

the vc model wants you all-in on one thing. but resilience looks like running 14 experiments at once, not one startup praying for product-market fit.

portfolios are the new moat. single saas is the new risk.