- find your niche and start from there - if you go too broad, you’ll die before you get adopted
- validate before you build - if nobody bites, the issue isn’t engineering
- focus on one GTM channel where your ICP already spends time - meet your people where they are at
- positioning is your moat: clear problem statement, clear ‘why this exists’, strong POV that will attract your ICP
Lovable
I’m sure all of you have heard about Lovable and most of you have tried it, but I’ve decided to start with one post about it anyway - not to teach you how to use it but to share the lessons it taught me.
Getting started is pretty simple: you create an account, type a prompt, and boom - full-stack app! A few more prompts and you are ready to ship, right?
Honestly, if you haven’t shipped your first unicorn by lunchtime, it feels like a wasted day! 🙄
Now, the reality is a bit different.
Your first prompt decides your fate - you will either end up with something useful or burn all your tokens running in circles.
Lesson learned: improve your prompting! Break tasks into small chunks, be specific about what you want to change, what it should do and what it shouldn’t touch (location+behavior+guardrails). This will help you with any LLM later on
It is (at least it used to be) very prone to primacy and recency bias - it remembers the first and the last thing you said.
Oh boy, how many times I had to repeat myself.. “use the same style for those buttons”.. “the SAME style, make them green”.. “green, not yellow.. green!”
Lesson learned: keep in mind it is always artificial but it’s not always intelligent 🥴 understand its limitations and the way of thinking, it will save you a lot of frustration - this, too, applies to most of AI out there
Somewhere during 2021/2022 (while Covid was a thing and Lovable didn’t even exist yet) I had a brilliant idea building a solution for restaurant/bar menus based on the QR codes.
Now, it already existed in most European countries but in a few markets there was an obvious gap - a million dollar idea, right?
Well.. we never got past the ideation phase. It would be pretty easy to build but, having full time jobs, the Fantastic Four (Ondrej Macko, Bojan Vasic, Ivan Tetkic and I) haven’t really found time to treat it more than a hobby project and… you know how the story ends.
Then, early 2024, me, Lovable and two afternoons - it was done! Landing page with all the restaurants/bars on the platform, individual pages (with custom subdomains and cute QR codes featuring their logos), admin panel to manage your menu (meals, drinks, desserts.. you name it). Ready to ship but.. the ship has sailed ⛵
Lesson learned: when building software is no longer a bottleneck, you need to double down on GTM!
My take on that: