ThinkFlow Introduction: The AI Development Experiment that turned into Production Ready App
Do you know the story where a monkey randomly hits keys on a typewriter and accidentally produces a Shakespearean sonnet?
The chances for that are… close to zero.
In another story - typewriting machine is my coding assistant. Or multiple assistants.
I’m not sure who the monkey is, but I am the one typing on the keyboard.
What came out is: ThinkFlow.
And about chances…
What are the chances that having a burning problem you want to solve, an army of coding assistants, not knowing the target language and technology, you get a production-ready application?
Yes, I almost summoned the word “vibe coding”, so let’s get things clear.
This is not a typical vibe coding story.
No “what I built over the weekend”.
No “SaaS built” by Gemini 3, or Claude Code.
Just a lot of “You are absolutely right” followed by reverting commits, tuning prompts, reviewing specs…
Before we get technical, let’s start with a problem.
Brave for personal email.
WhatsApp for communication.
Chrome for work-related stuff - I mean, something needs to host 120+ tabs.
Another Chrome in incognito for testing.
Two ITerms.
Three Warps. I am not mentioning tabs anymore.
At least five Visual Studio Code instances…
Slack, Todoist, Viber…
You get the feeling.
And then the dance starts:
“Where is my Code window for the bug I am currently working on?”
(swiping with two fingers on magic mouse…)
“Oh that sound is a new email, but where is my Brave…”
(making impossible keyboard shortcuts…)
Zoom call starts, I am pretending I have audio issues and blame it on Zoom until I find the screen I want to share, and hide private things…
…
Maybe I didn’t know where my IDE was, but I knew one thing for sure - I didn’t want to manage windows, minimize or hide apps, I wanted to focus quickly on work.
There’s no value in arranging apps.
And then I came up with an idea showing up behind the question…
Can I build the solution for my problem using AI coding assistants - in a language I’ve never used in production?
Can I apply everything I know and everything the best companies use when developing software?
(What are the chances to use “vibe coding” to get a running production-ready application, that doesn’t end up on Twitter and provide tokens for the rest of the hacker community.
My bet: deep software architecture knowledge matters more than syntax familiarity. If I can think in systems, the AI can handle the implementation details.
)
At least I can try. And share a story.
So here we are - three to four months of work planning and building ThinkFlow - true application workspace management for macOS, not just window positioning (but it does that too and many things more).
Space: the final frontier.
True for the Space, but for the ThinkFlow WorkSpaces this is just a beginning.
These will be my voyages with strange new technologies.
I will be exploring strange new prompts, seek how leadership and developers can use AI assistants better, and push my limits to boldly share everything: wins, failures, and unexpected lessons.