This is second in my fragment series where I share what all running in my life and mind. I do it for my own brain defragmentation.
A week that was nothing less than a roller-coaster ride. The highs were creating my first agent, completing a 40-hour fast, and getting better at Python. The lows were two days of being unwell, not being able to write a few emails I wanted to, and not having enough time or energy.
My father’s death anniversary was an intensely emotional moment. I wrote a private letter to him that I wish I had written a couple of years back, when he was still here. I realized that the guilt of not being able to express our emotions to our father can feel permanent. At the end of the day, the only solace is that, as my father, I am sure he would have understood my situation and forgiven me for what I could not do or say. I wish we had the wisdom to do or say what we are meant to, every time. Why is it that we realize we were wrong only after such a long time, when we can’t repair it?
With this emotional elephant out of my mind and on paper, I felt good celebrating Shivaratri this year. I did a fast that I had wanted to do for a long time. It starts at sunset the previous day and ends at sunrise the next day, which is roughly a 40-hour fast. I believe internal energy (a loose translation of prana) gets overused through metabolic activities. When we pause those metabolic activities, our prana has enough energy to be regulated and moved upward to align with cosmic energy. Here, internal energy is called Shiva and cosmic energy is called Shakti. Shiva moves upward through our primary seven nervous plexuses (chakras) when we sit with our spine straight and keep our mind calm and undistracted. This is the purpose of Shivaratri, and fortunately it fell on a Sunday, when I had the freedom to do nothing.
The mistake I made was that when the fast ended on Monday, I should have ramped up food intake slowly. Instead, I ate normally right away, and that caused a severe stomach upset. Anyway, lessons learned, and it still feels like a great accomplishment in my own terms.
My recent love for building coding agents got a real start with my first coding agent. I built a Python coding agent that takes a prompt in text form and writes complete, structured, unit-tested Python code using the Claude API. You can read about it here. It is a 30-minute exercise, and the initial investment is $5. Here is the article for your reading pleasure and if you can build one yourself then give my article a clap.
I really like the way Anthropic is progressing. They are fast, agile, and world-class. A recent article by Marco Kotrotsos—Anthropic Is Running a Different Race—explained well why they are doing so well. They are definitely winning. I just hope they can sustain this for the long run.
The 90-day company
Strip away all the analysis and one thing stands out. Anthropic plans in 90-day cycles. They ship products in 10 days. They run on vibes and improv and the death of ego. And they’re outpacing companies with ten times their headcount.
This isn’t just a cute cultural quirk. It’s a competitive weapon.
Speaking of Anthropic, I discovered their agent “play store” (or app store). It’s amazing: 100+ agents, skills, and templates in one place, and they are free to use. It’s a great place to start if you want to see how agents work. You can access it here.
I know Python, but I have never taken a formal course. Right now, I am learning slowly and steadily, and it reminds me of my initial days learning C, C++, and Java. It feels better to learn step by step than to just code, because it helps you understand the “why” behind many things. My companion for that is Tutedude.com. They have interesting schemes where you get your money back if you complete the course within a stipulated period.
Another interesting article that resonated with what’s being discussed in tech circles in Bangalore is Why the Smartest People in Tech Are Quietly Panicking Right Now. For your reading pleasure. This is just a reaffirmation to what Matt Shumer wrote in his article or rather putting the same content in a different way. But below sentence caught my attention
Never in human history have the architects of our future been in such a state of existential panic, while the rest of the world casually complains that “the chatbot still can’t do math.”
I also got to consult a few friends on how to start their AI journey.
The steps I suggest are:
- Learn the basics of AI tools like ChatGPT, Claude, or Gemini, and practice good prompt engineering.
(I am designing a prompt engineering course. I know there are thousands available (and free), but I am creating one for the pleasure of creating.) - Write about your learning and the benefits you observe, so you get community feedback and motivation.
- Build resources (AI resources like agents or your own flows) that make your life better, so that if someone asks why you talk about AI so much, you can show how it has helped you.
My friends find this useful, and I like the feeling too.
That’s a wrap for this fragment. See you all in the next one.