Besar @moveToMoonlight

i build companies & products. ml, recommendation engines, on-device ai. https://t.co/zOSsYTs197 | https://t.co/qAXHgISnJG | https://t.co/v9soj7DXMv Joined June 2024

Tweets

186
Followers

102
Following

88
Likes

406

Besar @moveToMoonlight

3 hours ago

@jackfriks 100%. The luck is downstream of the volume. You just made enough things to roll the dice more times than everyone who quit

0 0 0 18 0

View Details

message to follow4follow gurus: they think they're gaming the algorithm. you're not. the ranker reads engagement-per-follower, so 1000 dead mutuals make you look worse than 100 real ones. you're optimizing the exact metric that buries you instead of follow for follow, you should be worth following. boring answer, but it's the only one that compounds

0 0 0 21 0

View Details

Besar @moveToMoonlight

3 hours ago

@thejustinwelsh people look for the secret because the secret is free and the work is expensive

0 0 0 15 0

View Details

Besar @moveToMoonlight

3 hours ago

@gregisenberg 7. people who can turn a model into something your mom could use. i've watched a great recommendation engine die because the ux around it asked too much. the model is 20% of the job. making it disappear is the other 80% 

0 0 1 70 0

View Details

Besar @moveToMoonlight

3 hours ago

neither. you build proof first. "audience vs product" assumes people will care once you have either. they won't. what converts is evidence the thing works, one user with a result you can show. get one person a real outcome, then build the audience around that story and the product around that pattern. everything else is building in a vacuum.

1 0 1 41 0

View Details

Besar @moveToMoonlight

4 hours ago

and thoes branches don't stay hypothetical. you hedge, you verify, you give less rope, they feel it, and distrust quietly manufactures the unreliability it was guarding against. and that gets us to this tricky question: how do you ever fairly judge someone you hve already started to doubt, when your own caution is now part of their results?

0 0 0 115 0

View Details

Besar @moveToMoonlight

7 hours ago

How to out-build people twice as talented as you

0 0 0 34 0

View Details

Besar @moveToMoonlight

7 hours ago

the skill that separates people who ship from people who start is not discipline. it's the willingness to be bored. the last 20% of any product is tedious, unglamorous, and where everyone quits to go feel the dopamine of starting something new. learn to find the boring part interesting and you'll out-build people twice as talented.

0 0 0 29 0

View Details

Besar @moveToMoonlight

9 hours ago

@frmbrice probably just the usual: US first because it’s controllable, EU later because lawyers

1 0 1 18 0

View Details

Besar @moveToMoonlight

10 hours ago

11 hours ago

The opposite is true imo, AI didn’t make selling harder. It exposed that most products were never worth selling. When building was expensive, shipping something was a moat. Now that anyone can ship, the only moat left is actually solving a painful problem.

Arvid Kahl @arvidkahl

a day ago

Even before AI, the hard part was always getting people to look at your product, let alone pay for it. AI, by making product work even easier, made selling even harder.

48 3 106 4K 4

0 0 0 36 0

View Details

Besar @moveToMoonlight

11 hours ago

@rauchg The insteresting part is that markdown wins precisely because it has no syntax to fight with. Anyone who can write a clear doc can now ship an agent. instructions.md as the new main() is going to age really well.