Sic semper tyrranis

How do you like living in the post AGI world?

OpenAI pro is so good at reasoning.

Prometheus, the god of AI art

with AI coding agents like Replit, API wrappers have no moat

The current sentiment around AI is largely caused by defensive reactivity, that arises from ontological shock when AI suddenly crosses boundaries that people were unprepared for.

In the future most of you will regret not burning enough GPT tokens in 2024. In exchange you produce synthetic reasoning, aka cognitive labor that is cheap and scalable. It's time to hire a prompt engineer.

Bluesky has 23 millions of users and only 8 of them e/acc? Why is this platform so decel?

The most impactful action at this stage is to burn more tokens on synthetic reasoning over all types of data. What's preventing you from benefiting from scalable synthetic reasoning?

the amount of money people are willing to pay for agents I'm building with Replit in minutes is unbelievable

web 4.0 will stand on the shoulders of AI agents

2025: most business challenges are a prompt engineering problem.

In my two-month experience using a Replit agent every day, I've gained several insights as a non-coder:

1. I found it helpful to prepare thorough instructions, a 'seed prompt' beforehand. This process was most successful, when collaborating with ChatGPT o1 on designing a seed prompt.

Prompt engineering often boils down to articulating your goal clearly and unambiguously

My top 3 tips for using Replit Agent as a non-coder:

1/3 Spend more time and effort crafting the initial prompt. Collaborate with the smartest AI on it.

Replit agent is the best AI no-code tool currently available. Success leans heavily on prompt engineering. Your seed prompt defines the course, and the agent is hard to steer. There is a lot of prompt engineering to learn, which is a fantastic opportunity to compete in a still blue ocean.