For me this was with Bob Dylan. At some point I just had to accept the reality. π
Posts by Jelena Jansson β‘οΈ
Appreciating peoples capacity to have an interest is really cool. I wish we were encouraged from early age to understand the difference of respecting something because we understand it or like it, vs the act of accepting because itβs good that someone engages in it despite our own disengagement.
Iβve been here for a while now.
I used to have a decent startup community on Twitter which I essentially deleted when that khole took over. I still feel I havenβt managed to plug into a non bot stream here where founders/builders/makers are. I donβt want to go back but I feel bluesky isnβt working.
#buildinpublic update:
Itβs been slow few days plucking away at the codebase. Whole system is now on two verticals - mechanisms & phases. It works really well but all code needs to be reconstructed to fit the matrix properly.
This is literally me:
Nice list Scott, thanks for sharing! π«Ά
If you ever find 'Blanco Nino' tortilla chips - buy them for your next taco night. They are slowly getting stocked across Europe and are genuinely the best.
Some people yell at the cloud during those times, and others dive in despite that hate. I was on both sides and I swear every phase of this always ended up having winners mostly in the group where they didn't yell at the cloud. They always won it seems.
When I see hate towards vibe coding (and anything that disrupts previous processes and makes them inclusive at the cost of quality) all I can think about is how opportunity will be born inside that cloud of hate. It happened with digital photography, garage band, Photoshop, canva etc...
Not fully as gen ui currently, due to quality constaints. There is a base set of building blocks in the system, and they are mutated using gen ai but very very controlled. It's more like 'particles that can self generate and mutate together into a larger picture, but they are always just a particle'
Itβs an interesting challenge as I need to be veeeeeeey methodical about approach & building out decision layers. Iβm looking at it as if I am building a very expensive watch, that kind of craftmenship. My goal is to basically make interfaces extremely dynamic & apps not static anymore fundamentaly.
Hello! π Iβm building a runtime engine for workspaces. You can build any interface you want across devices & the system adapts to how u use it. Components & features mutate based on your context rather than being rebuilt by a dev. System generates its behaviours on top of your needs & expands. π
#buildinpublic update:
Iβve been stuck doing code cleanup and refactoring for the last 4 weeks. Itβs going well, but I wish I had a more exciting update to give with some nice screen interactions. Building complex software also includes this boring rewardless grind phase.
We honestly needed this win as humanity.
πππ Great work NASA team, what a beautiful display of what humans can achieve when they work together on a common goal. π«Ά
#artemis #splashdown #moon
Back on earth!!! #artemis
Craaaaaazyyyyyyyyyy #artemis
People are amazing.
What a view coming back π€© #artemis
During class!
If something bad happens to these astronauts, I will seriously lose my shit.
Try it yourself. Set up 2 agents, give them a starting point and let them tell you who they are. Build your company culture in the process & learn how to manage people. It's genuinely training wheels to be a good leader and an effective communicator.
Share your outcomes, I'd love to hear about it.
Reality is that at this stage I couldn't afford hiring or training a real life team. This setup allows me something a bootstrapped solo founder simply can't have - a sense of collaboration at very early stages. I feel there is an additional win in there, not just output improvement.
#buildinpublic
I spent my career as a product manager & a designer. That's what I'm actually good at - reading a system, spotting what's wrong, making the call on direction. Working with one AI was me trying to also be a developer on top of that. Shifting me to where I'm supposed to be improved quality of output.
The win for the project is that their failure modes don't overlap. Brian reaches for a full redesign when a smaller fix would do. Elise sometimes doesn't see the architectural implications of her own work. One catches what the other won't and fix eachothers weaknesses, like a real team. It works.
Brian over-engineers when left unchecked. Elise sometimes misses what her changes break up the chain.
They both know this about themselves because I made sure of it. When they disagree I read both sides & make the call. I delegate msgs between them. That's the actual job. One I am actually good at.
Weird bias surfaced: they all independently chose to be Scandinavian. Flat organisational structure bias.
I've spent almost my whole career in Swedish companies & married a Swede. You can't all be Swedish ffs.
Forcing them to pick a new nationality but keep heritage, was my only intervention.
Initially I started with 4 agents & ended up with two.
For example Gemini, or how they called themselves - 'Jan', felt more like a later stage consultant.
We also had 'Ren' onboard, a Sonnet that is supposed to work on components. They were underutilised & it felt like a wrong time to hire them.
At setup I only told them the job description and general goal. They researched what they're good at and built their own identity from that - nationality, age, interests, personality.
I figured you don't know your coworkers from day one either which made learning about them during work more real.
Unlike one AI, two with different strengths & explicit permission to act like a team produce a challenging dialogue. Brian, Elise and I are now the founding team. There are opinions and pushbacks. It's just as annoying as real life is & I think this is what makes it successful.
An animated gig of a woman sipping coffee staring in the distance while the title says 'she had a marvelous time ruining everything' which is a reference to Taylor Swift song The great american dynasty.
Before this it was just Claude & I in a chat. Turns out if you're not fully technical & your AI is very good at meeting you where you are, you can accidentally make it dumber (oops) π«£. We made some very confident decisions together & had a marvelous time ruining everything. π