Twitxy Saved

Kinder • Grinder

Verified account This account is verified. Learn more

@kinder_grinder

🤖 Rebuilding @automatio_ai
🎯 SaaS backlinks @ vanila.io
🎬Make gifs on 8gifs.com
🚢Ship on @candevsdosome
🧪 Experiments @moonlyNFT, 🎭 PFP @karamendos
Milky Way automatio.ai Joined April 2022
1.9K Following 2.8K Followers 7K Posts

Latest posts 5

Pinned post

Kinder • Grinder Verified account @kinder_grinder
Hey folks, I'm Stefan, founder of @automatio_ai , a no-code web automation tool.

Right now, I'm on a mission to rebuild the entire platform from the ground up. I won’t be writing any code myself, but I’ll be using AI to help me do it. I’ve been in this space for 15 years, and while I’m not a developer, I understand the tech, I know how things work, and I’ve spent years testing, debugging, and designing logic and user experiences.

Currently, Automatio has a little over 50,000 registered users, but because of technical limitations, platform complexity, lack of mobile/browser support, and high churn, we’re sitting at a very low $2.2k MRR. That needs to change.

Why rebuild?

The current version is too complicated. It requires a Chrome extension, works only on desktop, doesn’t support mobile or other browsers, and there’s a big learning curve. Even then, it’s not always stable. That’s been bothering me for a long time. I’ve held back from talking about it publicly because I was never 100% confident it would work smoothly for new users.

So now I’m rebuilding everything. From scratch. This time, I’ll do it with AI helping me write the code, while I stay focused on logic, testing, and UX.

A bit of backstory

Automatio started as an internal tool back in 2017.
In 2021, I finally started offering it as a monthly subscription product. Shortly after, I lost my dev team. I went solo and started building in public. That’s when I managed to grow it to $5.5k MRR by myself.

In total, Automatio has earned close to $200,000, even with all the ups and downs.

But then I went deep into the blockchain and crypto space for a couple of years. Didn’t touch Automatio. Just survived the crypto winter.

Now I’m back.
In February this year, I officially dismissed the team because I couldn’t afford to keep supporting them. Since then, I’ve been working solo. I’ve spent the last 7–8 months experimenting, learning how to "white-code" with AI, and understanding how to use tools like Cursor to get AI to build what I want.

Because of my background working with devs for years, I know what they need, how they think, and how to prepare things for smooth implementation. Now I’m using that experience to guide AI through the building process.

What’s the vision now?

Automatio will become a ChatGPT-style assistant for web automation.

Instead of clicking around or learning workflows, you’ll just say:

"Go to this site and extract all the data."
"Fill out these job application forms for me using my resume."
"Log into my dashboard and download invoices."

No browser extensions. No steep learning curves. No coding. Just simple instructions and fast results.

What’s next?

I’m rebuilding the platform solo with AI. No more depending on developers or freelancers. I’ll guide everything — from logic to features to edge case handling — and AI will handle the code.

The goal is to launch the new version in the next 30 days, at least privately for early testers. From there, the goal is to hit $5k MRR, then push beyond $10k MRR as we deliver more value.

If you’re into automation, AI, no-code, or just like following founder stories, stick around.
This is me going all in again.
Let’s see where this goes.

Peace,
14 replies 1 reposts 59 likes 6.4K views View on X
Kinder • Grinder Verified account @kinder_grinder
Simple version of Lenny ants world simulation. Made from his tweet, 1 prompt, DS 4.1 flash with Automatio.ai harness.

This might be my new desktop background.

Link to live site in comments
0:33
Lenny Enderle Verified account @lenny_enderle 5d
No ant in this video was told what to do.

I built a fun experiment, an ant farm where every ant has its own “brain”: a decision model that sees only what that single ant can smell, touch and remember, and answers with odds for 15 real ant behaviours. The ant picks from the odds.

The body and the rules of the world come from ~80 papers on one species, read and built by Claude Fable 5.2.

The tunnels come from them.

Ants are the easy case. Nothing in this setup is specific to ants: describe what an animal can sense, list what it can do, and let a model choose.
View quoted post on X