Good conversations are public.

Recent activity & thoughts while building

Been working on Crowdwide again, and lately I've been thinking less about what feature should I build next? and more about why would someone actually use it?

Recently I improved some of the UI and made the newsletter system more interesting. It can now use someone's interests to show more relevant articles, posts and communities instead of just throwing random stuff at them.

I also added a public place where anyone can look around and get a taste of what people are actually posting before deciding to join.

honestly...

• You can build 50 features.
• You can make everything look nice.
• You can add hashtags, communities, recommendations, newsletters, search and whatever else.
• And the user can still go: "Cool." 😭

Then leave.

So the bigger question I'm stuck on is:

Why would someone choose this platform?

Is it for discovering things? Building a community? Sharing work? Meeting people?

Or are people just here to waste time anyway? 😂

Because people literally change platforms just to waste time on a different platform.

And then there's the creator side.

I've started thinking that creators need some kind of reason to keep creating regularly.

Maybe that's money.

Maybe recognition.

Maybe growth.

Maybe all three.

I don't know if that's universally true yet. It's just what I've started noticing while building and watching how platforms work.

That's also why I'm exploring possible monetization partners for Crowdwide. I've contacted a few companies about it, but so far... silence.

Still experimenting.

Still building.

Still trying to figure out what actually makes a platform worth coming back to.

Maybe that's the real FAAAAAH of being an indie dev.

Building the thing is one problem. Getting someone to care about it is another.

#buildinpublic #indiedev #opensource #socialmedia #crowdwide

🦉 From Broken Models to Living Systems

I didn't start with a perfect AI setup.

No GPU. No funding. Just ideas, experiments, and a stubborn urge to build.

These are old projects now, but each taught me something.

🚀 Lynqbit — My Favorite Failure

Lynqbit was my first serious AI project.

A roughly 90M-parameter model built with ambitious goals and plenty of experimentation.

Eventually, my system couldn't properly support the training process.

Two months of work were gone.

But the failure taught me something:

Training shouldn't depend on one fragile setup.

🦉 Barn Owl AI

Next came Barn Owl AI.

I experimented with streamed training, cloud-hosted data, and incremental learning.

It didn't last long. Infrastructure problems and bugs eventually stopped the project.

The project failed.

The idea didn't.

🦉 Elf Owl AI

Then came Elf Owl AI.

This became my first AI experiment that actually reached a usable state.

• 25M parameters
• Custom training data
• Open-source
• Publicly released

It wasn't perfect.

It hallucinated, struggled with grammar, and had plenty of quirks.

But it worked.

That felt like a huge victory.

🦉 Xenoglaux AI

Later came Xenoglaux AI, another experiment focused on streamed and modular training.

The dataset was larger, but the same enemy remained:

Hardware.

Training that took hours on a GPU could become painfully slow on CPU.

🧠 What I Learned

• Failure is compressed knowledge.
• Small models can still be interesting.
• Hardware limitations force creativity.
• Every failed experiment can improve the next one.

These projects are finished chapters now.

But their lessons stayed with me.

Sometimes the model fails. The idea doesn't. 🦉

#AI #MachineLearning #Python #OpenSource #Programming #BuildInPublic

🦉 Elf Owl AI — The Little AI That Could

“It’s not about perfection. It’s about vibes.”

A few years ago, I experimented with building a small AI from scratch.

That experiment became Elf Owl AI.

🌌 The Little Owl

While huge AI models were using billions of parameters and massive computing power, Elf Owl followed a different idea.

It was small, handcrafted, and experimental.

• Around 25M parameters
• A custom handcrafted dataset
• Built with Python and PyTorch
• Designed with an owl-like personality

I wasn't trying to compete with massive models.

I wanted Elf Owl to simply be itself.

It had unusual responses, imperfect grammar, and plenty of quirks.

Sometimes it made mistakes.

Sometimes it produced surprisingly interesting responses.

Those imperfections were part of its charm.

⚙️ The Hard Part

Running it was another story.

The model needed more resources than I could comfortably provide, and keeping it hosted was difficult with limited computing power.

Eventually, the project became another experiment I had to leave behind.

🧠 What I Learned

Elf Owl taught me that building AI isn't only about making the biggest model possible.

It's also about:

• Creativity
• Experimentation
• Working within limitations
• Giving software its own character

I once imagined building Elf Owl v2, v3, and beyond.

That never became the reality I imagined, but the idea taught me plenty.

🦉 Looking Back

Elf Owl AI is an old project now, but it's still one of my favorite experiments.

It wasn't the smartest.

It wasn't the biggest.

But it had something I cared about:

Personality.

Sometimes, that's enough to make a tiny project memorable.

#AI #Python #MachineLearning #Programming #OpenSource #BuildInPublic

🐱🦉 From Lynqbit to Elf Owl AI

Sometimes, looking back at old projects shows you how much you've actually learned.

A few years ago, I experimented with building AI from scratch. These projects are long finished now, but they became important chapters in my journey.

🐱 Lynqbit — My Digital Cat

Lynqbit was my first serious AI experiment.

I built her with PyTorch and spent months experimenting with training, data, and personality.

I wanted her to feel different from a normal chatbot. She could give playful answers and behave like a cat.

Eventually, Lynqbit reached around 90M parameters, and my hardware couldn't keep up.

The project ended.

But it taught me something important:

A failed project can still be a successful learning experience.

🦉 Barn Owl AI

After Lynqbit, I started Barn Owl AI.

I wanted to create something more thoughtful — an AI inspired by the idea of a wise owl.

The project was eventually paused and abandoned as I moved toward other things.

But the idea stayed with me.

🦉 Elf Owl AI

Later, I experimented with Elf Owl AI, a much smaller project.

• 2.8M parameters
• Around 7MB of curated data
• Built with PyTorch

It was tiny, imperfect, and sometimes crashed.

But it taught me plenty.

💭 Looking Back

These projects are years old now.

I don't maintain them anymore, but their lessons stayed with me:

• Start with what you have.
• Failures are part of learning.
• Small experiments can lead to bigger ideas.

These were old projects, but they helped shape the developer I became.

#AI #Python #MachineLearning #Programming #BuildInPublic #ArtificialIntelligence

🐈💀 When My AIs Died: The Rise and Fall of Lynqbit & BarnOwl

Not all code runs forever. Some of it purrs once... then fades away.

A while ago, I decided to build my own AI systems from scratch.

Two of them became especially memorable: Lynqbit and BarnOwl.

🐈 Lynqbit AI — The Cat That Learned to Think

Lynqbit was my first serious AI experiment.

I wanted it to have personality, so it didn't just answer questions — it acted like a cat.

• Built with PyTorch
• Custom BPE tokenizer
• Around 94 million parameters
• Trained using my own dataset of 57k+ questions

After roughly two months, it started producing surprisingly good responses.

Then reality hit.

The model became too heavy for my available storage and computing power. Training slowed down, databases struggled, and eventually I had to let Lynqbit go.

RIP Lynqbit. 🐈💔

But it taught me that even a failed experiment can completely change how you think about building.

🦉 BarnOwl AI — The Night Scholar

I wasn't finished.

BarnOwl was supposed to be different — smaller, quieter, and more thoughtful.

I wanted it to observe and reflect rather than simply throw information at people.

For a while, it worked.

Then came the resource problems.

Training repeatedly crashed, datasets disappeared, and my hardware simply couldn't handle what I was trying to build.

After countless retries, I had to accept it:

BarnOwl couldn't fly. 🦉💀

💭 What They Taught Me

These projects taught me that AI isn't just algorithms.

It's compute, planning, patience, resources, and persistence.

Both AIs died.

But the lessons survived.

And honestly?

I'd build them again.

#AI #Python #Programming #MachineLearning #BuildInPublic #ArtificialIntelligence

🚀 Building LynqBit: The Story of My First AI Project

LynqBit was one of my older projects, built while I was still learning programming, AI, and what I could actually create on my own. I wanted to share its story as part of my build-in-public journey.

⚡ Where It Started

Like many developers, I had experienced AI confidently giving completely wrong answers.

Instead of only complaining about it, I started wondering:

“Why can't I build my own AI system?”

That question became the starting point for LynqBit.

I spent countless hours going through tutorials, documentation, experiments, and debugging sessions.

There were API limits, restrictions, paywalls, and plenty of things I couldn't easily experiment with.

But that frustration pushed me to learn more.

🔥 What I Built

Eventually, LynqBit was born.

It wasn't a massive AI company or a polished commercial product.

It was mine.

• It eventually answered 56k+ questions
• I could experiment with its logic
• I could break things and rebuild them
• Every failure taught me something new

The first time it correctly answered something using logic I had written myself felt incredible.

It wasn't just code working.

It was proof that I could actually build something like this.

🚀 What It Taught Me

LynqBit didn't become my final project, and that's okay.

It taught me that I didn't need to know everything before starting.

I could learn while building.

I could fail, rebuild, and keep going.

Most importantly, it gave me the confidence to take on bigger projects.

LynqBit was one of my first big experiments.

And now, I'm carrying everything I learned from it into what I'm building today.

This is an old project, but its story is still an important part of my journey.

#AI #Python #Programming #MachineLearning #BuildInPublic

Teaching Users a Feature Is Surprisingly Hard 😅

One thing I’ve noticed while building my projects: creating a feature is sometimes easier than getting users to actually use it.

For example, in one of my projects, Rizzzler, I added an option to put music on your page. It’s completely free, with no ads or promotions, and I even added it to the documentation.

Out of 10+ users, only 2 have tried it. I even shared my own page to show how cool it can look… and still, most people stick with the default music. 😂

And Crowdwide is teaching me the exact same lesson.

I’ve added features like:

• Hashtags for posts, articles and polls
• Community posting instead of only posting from your personal account
• Scheduled posts
• Quests and co-authoring
• Recaps
• A world map where communities can show their location 🌍

But most people simply write their text and post normally. 😭

The funny part? I’ve personally explained some of these features and even messaged people about them. Still, some features remain untouched.

The map currently has basically one resident: my own community. 😂🌍

So maybe the problem isn’t “users are lazy.”

Maybe, as builders, we’re not doing a good enough job of showing users why a feature matters.

Documentation alone might not be enough. Sometimes users need to discover a feature naturally, understand its benefit, and have a reason to try it.

Building the feature is only half the job.

Teaching people that it exists — and making them want to use it — is the other half. 🚀

#BuildInPublic #Crowdwide #IndieHacker #UX #ProductDevelopment #WebDevelopment #SaaS