Founder Recap Apr 2026
Founder Recap Apr 2026
Feels like the world kind of shifted again this past few months, my brain is suffering from indigestion with how fast things are changing and I thought one of my skills was being good at drinking from a firehose of information but I got swallowed by the tsunami. I’m sad to say that progress with Rekait has been slow recently, I wanted to pursue new directions with things, to be more focus on niches and directions that I felt I had a good chance as a solo founder. But lately I’ve been spending more time building things that are not core features for Rekait which is really not good because momentum feels so important to keep things going.
I’m not sure yet whether the recent detours will be worth it in the long run but if I keep hearing myself say I wish I had done this sooner or built this sooner, that I wished I had this earlier, then probably I won’t regret it if I already regretted not having it before. It feels like a really weird and difficult balance to strike. On one hand, I felt good that I was focused on Rekait and just be obsessed with what it could be, on the other, I felt like I wasn’t keeping an open enough mind for some of the smaller things that could add up towards the bigger picture.
One of the more recent realizations is I always thought I was pretty good at using LLMs to write “good” quality code, but that doesn’t necessarily mean that I’m really efficient at using LLMs to write “good enough” quality code. My most recent mindset shift is asking myself more the question of whether I should write a script or write a markdown file for a task, LLMs are just getting so good at following instructions (usually). Balancing the choice is an interestingly difficult thing to do because not only do you need to have a sense of where the edges of an LLM’s capabilities are, you also need to check if a certain model provider’s inference and harness caught a flu today or not.
But I still think core foundations for important software deserves high scrutiny and I still want to make sure I have a good mental model of the core abstractions, but there’s so much stuff in life that should probably just be markdown files handled by coding agents.
This is a long post to explain, nope I’m not winning yet.
