Ad guy turned builder. Working on 🪐 https://nitter.gq/t.co/q539seTVEO and writing down what I learn as I go: the product, the numbers, the mistakes.
Reality
Joined January 2022
- Tweets219
- Following150
- Followers165
- Likes635
Pinned Tweet
My job was to get inside the heads of millions of people.
Strategist at Publicis. Netflix, Heineken, McDonald’s.
From the inside, I saw how often big agencies sell more than they deliver.
So I left to build my own things.
3 startups didn’t make it.
Brands are buying AI visibility scores they can't rely on.
I'm concerned they'll make product decisions with confidence their tools haven't earned.
They believe they're in control.
Using an unreliable score, a brand manager risks approving listing changes without knowing whether AI visibility improved.
To me, that false confidence is more dangerous than knowing you don't have a measurement.
Everyone is sharing these huge stacks of tools and agents to run a business on "auto-pilot". I think they're wrong.
Believe me, I was a very early AI adopter, but I feel like it's become more confusing than it helps lately.
What's the most simple (AI) tools stack you use? My top 4 ↓
1. @meetgranola for meetings, note taking and anything I need to remember.
2. @bot as my personal assistant
3. @Claude Code for content
5. @gojiberryai for lead generation
(the picture is an example of what i find so overwhelming - found it on LinkedIn)
Our engineers had bigger features to build than Excel exports.
My marketing background helped me see why the spreadsheet deserved attention.
I've taught myself to code.
For Cosmy, I coded the landing page and blog, and automated the newsletter.
Outside that work, I've built apps for myself and use AI daily to write code.
show up everyday, even when it feels tough and lonely - consistency always wins in the long run.
let's go builders! ✌️
You can't trust Amazon listing scores from a single prompt.
For brands using CoSMo or Alexa for Shopping scores, that leaves a basic measurement problem.
Then we defined detailed criteria for every score level, with gold standards to check against.
That's the work behind Cosmy's scores.
I need a defined scale before trusting a number from a model.