Inklings

Anthropic says Claude now writes more than 80% of its own code, and names the next step: recursive self-improvement. I'd felt the shape of this in my own work for months without a word for it — exhilaration and unease, side by side.

Daniel Pink lists sixteen skills every functional adult should know. I watched it as a leader of people — and the list of how to be a grown-up turned out to be almost the same list as how to lead. Nobody taught me either.

My default is action — get it done, reward the doing. Sitting with a quieter dimension inspired by Chandra, the moon, and taught by my sister: how do I become a better nurturer, of myself and the people around me?

The world rewards being the best when everyone is watching. Yogi wisdom asks the opposite — can you be number one when the only person watching is you?

An Airbnb PM writes a love letter to the Google Chrome team about all-blue tab icons. A Google PM replies with the perfect comeback. Two brands, on each other's side, in public.

Speak only when you feel that your words are better than your silence.

When I look at myself — or at someone else — am I seeing a bundle of problems, or am I seeing potential? Naming the shadow, and asking whether the other lens can be practised.

A family, six lakh rupees, and an MCA decision I can't yet give. My instinct, the question I put to my network, and the advice converging from what came back — a page I'll keep updating.

Years ago Vikram told me we learn in layers. Venky calls it the power of compounding. Watching Claude connect the small tools I built over many years, I'm finally learning both — through AI system architecture.

As I think about AI, it's not about tools or process. As a leader, my thought to execution speed is now in minutes and hours, not weeks or months. That's the real shift IMO.

Playing a founder in a senior compensation conversation taught me something: operational stability is table stakes. The hook was never the salary. It was the story.

Peliqan is operational context infrastructure for AI agents — a cache, transformation, and inference layer that sits between Claude and your live business systems. I'm working out what that gap actually is.