There are many possible answers:
Kardbrd.com started as a fun way to learn and improve myself in my craft. I had some tokens and time to spare and good mood to vibe-code something for the soul. So a family kanban board that is shown on TV and makes everyone activitiy planning fun would be a nice addition to holiday moods.
Little I did know how it turns out. Initial prototype was done in minutes with prompting. Kanban board is a pretty well-defined structure that was well documented before. Next tokens are verypredictable. For every feature I've put a card and moved it from left to right, one by one. Felt like I was 5 years old again and was "playing work" with my toys. Pure joy.
Then came this idea that I can just create the cards, and because of the repetitive nature of my approach, vibe-coder tool would be able to pick up the card from the board, see what needs to be done, do it and send me for review. Things were becoming more serious. My friends just sent me links [beads](https://github.com/steveyegge/beads "A memory upgrade for your coding agent"), [vibe-kanban](https://www.vibekanban.com/ "Vibe Kanban lets you run coding agents in parallel without conflicts, and perform code review through our diff tool. Now you can focus on planning and quality instead of watching terminal logs."), [agenticcoding.ai](https://agenticoding.ai/ "Structured methodology proven on enterprise mono-repos with millions of lines of code"), but I was too busy with my cards and ideas to look around.
When I actually looked I saw that my idea is already validated, discussed, explored, open-sourced, but not launched. So I transformed my frustration with LLMs into vision and later Kardbrd.com. Using the same methodologies I've described on index.
I wanted the cards to move by themselves. With me just adding them and making checkpoints. Ideally using mobile screen. it worked! So I'm offering you to try it as well.
for there's no other material.
LLMs are the crime of the century. Nobody gave explicit permissions for their work to be used for training any of the LLM models. At the same time the genie is out of the bottle. We can ignore it, we can avoid it, but we can't put it back.
Eventually LLMs will become local, available, effective. I don't think adding more RAM and GPU will help the models better understand the context of the task they are delegated to execute. Precision over one-shotting. Context contents matters more than context window. Imagine your ideas being developed beyond the last byte of context.
If corporations don't need people to do the job
Kardbrd.com is the first step in that direction. Common space for you to work both with humans and LLM agents. Calm, async defaults with seemless sync on the background.
Summarizing everything said my why was bold: to build a space that will employ LLMs for actual work, organize my ideas and share
them with my coleagues and friends, collaborate without stepping on each other toes.
LLM inference is cheap right now, corps are subsidizing it to get people on the needle. Ideas were cheap, now the implementation is even cheaper. But it will not stay like that for much longer: shareholders will ask for the dollars they invested in these datacenters and H200 (or whatever is the current generation of shovels by NVIDIA) to be returned back and then those who got hooked up will pay a pretty penny.
So in future, we will be competing with peers on the marketplace of ideas. Those ideas don't belong to anyone, they are in the air. Anyone can materialize them any time, just have the right mindset and intentions. I've built Kardbrd.com to store my ideas and my offer is to store your ideas and make them accessible to you any time.
Why should I pay you? I can vibe-code this myself.
Yes! I'm glad you've asked. Let me help you a little: I've built this in over a month spending $100 with Anthropic. You will be able to clone this probably even for less time and money. Please help yourself. If you make significant progress and make it available for others then this project turned successful -- write me a note, I would love to know how you did it. The price I charge is a modest fee for hosting, maintenance and further iterations on the idea. So that you didn't have to waste you tokens and time.