essay
- Writing a message to a specific audience while taking into account a message history or context (I use Claude Projects for that)
- Generate ideas to solve a problem—although I try to do that less often to limit my resistance to thinking, which is alarmingly increasing
- Analyze data (PDFs, spreadsheets)
- Help me identify reasoning or logical fallacies in my writing
- Learn about new concepts
- Investigate trends
- Answer cooking and household questions
- Analyze personal health data (privacy hazard, of course)
- Provide high-level legal advice
- Perform comprehensive search for a given subject, etc.
- Keep conversations below 200k tokens (about 50k words). I don't believe in this 1M token thing (I may be irrational)
- Edit an already sent message instead of trying to tell the AI to correct itself if its response was bad. So let’s say I ask it a question about solar power engineering… it answers in an academic way but I wanted a business approach… instead of sending a new message saying “i want this to be about business”, I edit the first message I sent to steer the AI (less tokens burnt per day, and answers better, as any noise in the context window will prevent optimal behavior)
- Improve prompts when asking about mission critical stuff (assumption: humans don’t know how to prompt). I use the Anthropic Console for that or I use speech-to-text. I noticed that longer prompts are better and it's easier to speak than type
- Be diligent about Project context (as little noise as possible, delete old files, clean up regularly)
Why speed matters for early stage companies
A simple intuition is that it is about course-correcting.
Velocity mitigates the mistakes in positioning, especially early on. You don't have to be fast. It's just that a team that executes faster will mechanically be able to “try again” more times than a slower one.
How I use LLMs for work and life (Spring 2025)
I thought I'd write a short post for future reference.
This is how I use LLMs daily for different tasks, such as:
I mostly use o3 and Claude 3.7 Sonnet (Extended Thinking).
Some heuristics:
How software mediates thought and constrains free will
There is a fundamental difference between modes of composition: dictation, pen on paper, or keyboard. These differences noticeably influence the final output, although the effects remain unpredictable, emerging downstream from the complex system of communication.
Sometimes I'm overly aware of the chosen mode and find myself asking: what would the output—and thus the communicated message—have been if a different mode had been selected? What would Socrates' ideas have become without Plato’s writing? How would Kant’s thoughts differ had he used a keyboard? And what if Aaron Swartz had relied on pen and paper?
Even when expressing the exact same idea—if such a thing is possible—the resulting outputs might differ so profoundly that the behavioral outcomes become even more unpredictable. Different modes of composition inherently create different outcomes, influencing the subject’s possibility of self-determination in digital environments.
✴✴✴
Digital composition is uniquely distinct because its interfaces are intentionally designed. They embed someone else's mental model of creativity and thus limit possibilities, exerting a constituting power of creating contingent “reality” and possibilities—or presenting composition merely as one option among many decided by the designer.
Physical tools (paintbrush, pen), by contrast, have constraints arising from material reality rather than another person's mental model, fostering a distinctly different relationship to free will.
Marxist inversion: software is marketed as “tools”—the greatest farce—yet they lack the essential nature of true tools, which inherently offer maximal free will within their physical constraints.
The homogenization of writing technology (keyboards, word processors) may consequently limit the emergence of novel philosophical ideas. Modern UIs function as non-neutral mediators of thought—an invisible architecture shaping our thinking beneath our conscious awareness.
Perhaps this is why we cannot yet adequately theorize social media. Understanding the influence of "tool software" on composition, communication, and free will is a prerequisite to unraveling deeper “social software” aspects: feedback loops between composition and response, algorithmic effects, and the creation of new social norms.