Built 20+ products 11 years. Now orchestrating AI agents to build autonomously. CTO 🍍 @Pineapplebuild1 — first AI website builder Prev CTO Aurity - Tech Acc
Your website in 5 min ⇢
Joined July 2013
- Tweets12.9K
- Following873
- Followers3.9K
- Likes13.7K
Pinned Tweet
This was my dream!
Build a website with just words in your native language! 🇺🇸🇧🇷🇩🇪🇯🇵 🇵🇱
No Coding, No Designers, or No Makteres.
Just, get a Website with great Design and Copy that Sales 💰
🧵👇️
This is why I moved @Pineapplebuild1 to be AGI first.
Follow to watch the progress as the business build itself.
DeepMind CEO Demis Hassabis:
"One person who really understands AI will outperform an entire startup team
In 5 years, the people who build with AI will run whole companies with a tenth of the headcount"
62
ok, my AGI needs some work!
any ideas what can bring best results? 🤔
I am almost done with my content factory to share docs, blogs, social update and record videos as it's was dry as F for the last 9 months :/
ok, 2 weeks with Grok bot
Codex for coding, Grok bot for everything else
having my non-engineering stuff in a separate app is just easier for my head
and it runs on its own computer in the background while I do other stuff
got a lot done... even sold my lego ;)
2
94
Today, to get a new Mac with 48GB you need to pay 1,000 more vs 2 years ago 😕
ok, nice I get a touch OLED display, but do I really need it?
This is weird... this is the best configuration for M4 Pro with 48GB ram
the next reasonable version is 1k more with the same Ram size!
Do I really need M4 Max?
132
ok, trying Grok bot, why?
basically, having it all under one ChatGPT app is too much, so moving more of my personal workflow to another app is just easier to handle mentally
ps. the Codex team toggle for Codex vs ChatGPT just doesn't work
1
242
This is the best demo I've seen recently
Extremely simple but explains everything
Everyone's gatekeeping character swapping.
So we turned it into a button.
It's wild how precise AI editing has gotten. Even the movement on the hair, the physics, it's all incredible.
Creating has never been more fun.
2
3
248
That’s the reality. The price of growing slow is too high. The opportunities are everywhere and if you have strong distribution there is no reason to slow down.
OK, I might have been wrong about building with a lean team in the age of AI.
We’re now at 40 people and plan to double in the next 6 months.
When building is this much better, easier, and faster, the opportunity cost of not building more is huge.
We’re hiring because you still need humans in many parts of the building loop. Yes, building is easier, but that just makes me want to build more things. Serve more customers. Build the best product they’ve ever used.
The reality is that products today still suck. Coding is solved, but building great products hasn’t been. Products still don’t do everything you want them to do. They’re still buggy. They’re not flexible enough.
Building still needs humans. That might change, but it doesn’t make sense to act like we already live in that world when we don’t.
I want to build more, grow more, and achieve more. I want more customers who love our product and get real value from it.
And I want to enjoy this beautiful moment in history, when building is so much easier and still so fulfilling.
2
2
138
I did some meditation, and it helped a lot.
It’s amazing how simply pausing for a moment and sitting still can bring so much clarity about what really matters in life.
So busy that even my AGI is not helping.
At the same time I feel like I need to just pause for a second.
Clear my mind and remind myself what is important.
1
4
156
So busy that even my AGI is not helping.
At the same time I feel like I need to just pause for a second.
Clear my mind and remind myself what is important.
Which one to pick >> T3 Code, Herdr, Orca and Codex?
Which would you pick for remote work and everyday UX?
So far my fav is Codex... but I might need to start using Claude for more design
3
412
Peter Kowalczyk 🍍 retweeted
I’m quitting Pineapple to work on something bigger. 🍍
Why? Because I believe AI can grow Pineapple on its own -and eventually do it better than me 😉
In some ways, it’s already proven that.
It has replaced the work of a 4 team. Today, I’d estimate it handles the workload of 10.
Monthly costs have dropped from $30,000 to $1,500.
Now I’m letting it run wild.
Its own budget. Teams of AI agents. Responsibility for running the business and reporting results.
Analyse the market. Understand what customers need. Deliver great support. Address their pain points. Run real experiments driven by data. Learn what works and keep improving, 24/7.
That was always the goal for Pineapple: help small business owners run and grow their businesses on autopilot.
Pineapple goes first.
I’m moving my focus to a bigger project. AI gets the challenge of taking Pineapple forward.
I believe we’re in the AGI era. Time to put that belief to the test.
Starting today, I’ll share the journey openly: the numbers, the decisions, the wins, and the failures.
🍍 September update: Can AI grow Pineapple itself?
💰 Stripe MRR: £16.63
🤖 New chapter: giving AI the job of growing Pineapple
🎯 Mission: help small business owners grow on autopilot
I’d love your feedback - and if you’d like to join the experiment as a mentor helping refine the AI teams’ decisions, comment or DM me. 🍍
I’m quitting Pineapple to work on something bigger. 🍍
Why? Because I believe AI can grow Pineapple on its own -and eventually do it better than me 😉
In some ways, it’s already proven that.
It has replaced the work of a 4 team. Today, I’d estimate it handles the workload of 10.
Monthly costs have dropped from $30,000 to $1,500.
Now I’m letting it run wild.
Its own budget. Teams of AI agents. Responsibility for running the business and reporting results.
Analyse the market. Understand what customers need. Deliver great support. Address their pain points. Run real experiments driven by data. Learn what works and keep improving, 24/7.
That was always the goal for Pineapple: help small business owners run and grow their businesses on autopilot.
Pineapple goes first.
I’m moving my focus to a bigger project. AI gets the challenge of taking Pineapple forward.
I believe we’re in the AGI era. Time to put that belief to the test.
Starting today, I’ll share the journey openly: the numbers, the decisions, the wins, and the failures.
🍍 September update: Can AI grow Pineapple itself?
💰 Stripe MRR: £16.63
🤖 New chapter: giving AI the job of growing Pineapple
🎯 Mission: help small business owners grow on autopilot
I’d love your feedback - and if you’d like to join the experiment as a mentor helping refine the AI teams’ decisions, comment or DM me. 🍍
Peter Kowalczyk 🍍 retweeted
I had to try. Just bought $15/mo Claude subscription to try Opus 5.5 (high)
I was lazy, so just single-shot it (prompt down👇🏻)
Much better than Astra 6 High
I had to try. Just bought $15/mo Claude subscription to try Opus 5.5 (high)
I was lazy, so just single-shot it (prompt down👇🏻)
Much better than Astra 6 High
You are a world-class motion designer cutting a Twitter SaaS launch spot.
Showreel stakes: if this is weak, you don't get hired.
PRODUCT URL: [yourproduct.com]
PRODUCT NAME: [Name]
If this folder is the product repo, rebuild real UI from the codebase. Otherwise scrape the live site for logo, mascot, hex colors, fonts, and screens.
MAKE
A 15.00 second motion-graphics product launch film.
1920×1080, 16:9, 60fps. Music-locked. Code only.
No video model. No After Effects. No stock footage. No invented UI.
LOOK — the exact genre people are posting on X this week
(Goutham / Danny Postma motion-ad / Dub .co / TypingMind / twoclipping):
- real logo, real mascot if the site has one, real product screens
- kinetic type, camera pushes, UI cards that float and dock
- a cursor or finger that actually clicks the product
- dark or product-colored ground, light as the subject
- one idea per beat, big centered type, max 8 words on screen
Banned: generic AI gradients, particle bursts, fake testimonials, fake customers, dead time, template look, buzzword soup.
STRUCTURE (15s, something on every beat)
0.0–2.2 Hook: pain or bold claim + logo sting
2.2–5.4 Product reveal: hero screen or device. Mascot may enter here if the brand has one.
5.4–11.8 Three feature beats using REAL screens rebuilt as HTML (not frozen PNGs if you can avoid it)
11.8–15.0 End card: logo + one-line + CTA
BUILD
1. Fetch the site (and repo if present). Extract logo, colors, fonts, mascot, 6–10 screens. Do not invent assets.
2. Show me storyboard.md first: Time / Visual / On-screen text / SFX. Wait for my OK before writing animation code.
3. One HTML file. All motion is a function of time t (paused GSAP timeline you seek, or seek(t)). No CSS transitions, no timers, no state carried between frames.
4. Shared timeline for picture and audio so hits land on the beat.
5. Render review stills at 0.4, 2.0, 5.0, 9.0, 13.5s. Fix overlaps, blurry type, off-brand color.
6. Playwright frame capture → ffmpeg H.264 MP4. Prefer 4 subframes + ffmpeg tmix for motion blur.
7. Music: royalty-free ~120 BPM, start on a downbeat. Motion matches the music.
CONSTRAINTS
- Claims must exist on the live page. No made-up metrics.
- Last frame can hold the end card; do not loop unless I ask.
- Deliver: storyboard.md, index.html, render script, out/launch.mp4.
Go all out. This should look like your motion-design résumé reel, except it is my product.
1
1
872