Is slop a quality problem, or an intention problem?
Slop essays. Slop code. AI slop videos. Slop bowls for lunch.
I don't think “used AI” is always equivalent to “slop.”
At its core, I think slop is content where nobody made enough meaningful choices.
I've been thinking about this a lot because @trymelius helps people create images and videos using generative media models.
It's never been easier to ask AI to “make me a great video” without really thinking through:
- Who is it for?
- What should they feel?
- What's the message?
- What should the scene, script, lighting, characters, and composition actually be?
AI can make a lot of those choices for you. But without direction, it'll tend toward the average.
I can't believe as late as 2022 I was working at a tech company that only deployed/released software once or twice a week.
Then I joined @tryramp and saw that every code change I merged would be deployed within 30 minutes...
Shipping fast, learning, and iterating is the key for many software companies to compete and win.
Every day counts. Ship every day, multiple times per day.
We do @trymelius. 202 days and counting...
https://t.co/1LI8gUGlA8
I spent 3 years leading engineering teams @tryramp across Fraud, Risk, and Growth
Ramp taught me that regulated software doesn’t have to mean slow software. Be extremely careful with auth, permissions, and money movement. Move extremely fast everywhere else.
I left Ramp as coding agents were starting to actually work. It felt obvious that small, AI-native teams were going to build products and companies very differently.
I spent the next year testing that idea, including starting a fintech company with former Ramp coworkers. We could build incredibly fast, but convincing large asset managers to change how they worked was another story.
Now I'm working at @trymelius with @n0w00j, who was my first direct report at Ramp 😄
We’re building a platform for creatives to turn whatever they can imagine into images, video, and audio with generative models.
Generative media feels a lot like AI coding did a couple years ago: capabilities are moving incredibly fast, but getting exactly what you want is still hard.
Super excited for this journey, and I will share lessons learned as I go!