Digest 001
Things I'm reading, watching, and hearing
Way back when, in the halcyon days of a prior Substack that has been in a coma for ~4.5 years, I used some of the newsletters to share highlights from what I discovered in any given month (also music that I had on repeat).
While the idea of a regular digest causes PTSD, I’ve gotten a fair amount of requests for content that I recommend. Given that I find myself sending links out to friends and colleagues piecemeal and with alarming frequency, I thought that an occasional digest would be a fun experiment.
So, here we are. The format is simple: a link, and a short summary of why I found it worthy of my (and your) time. It’s up to you from there.
SemiAnalysis: Claude Code is the Inflection Point. Good read on Claude Code as the “ChatGPT moment” for agentic products. Regardless of whether you have used CC yourself, the piece offers a helpful glimpse into the current state of workflow automation and substantiates why Anthropic’s CEO believes that “50% of entry-level white collar jobs could be automated by 2030.” Claude Code has become an increasingly important part of my own workdays, from handling data analysis to building small apps for the Macoto team to developing the Gravity Shackles research assistant that helped curate these links. I had high hopes for the Cowork launch, but so far still find Claude Code substantially more useful.
Dario Amodei: The Adolescence of Technology. Speaking of Anthropic! This is the latest piece from Amodei (Anthropic CEO) and the counterpart to his 2024 article (Machines of Loving Grace). It explores the idea of a “country of geniuses in a datacenter” materializing in 2027 and then asks, “What should a national security advisor worry about,” in this reality? AI autonomy, misuse for mass destruction, misuse for seizing power, economic disruption, unknown effects from speed-running scientific progress, etc. Also implications for redistribution of wealth, regulation, and more. Light fare…
Anthropic: Claude’s Constitution. You might then want to read the text that Anthropic developed for Claude, in a thought-provoking exercise to govern its values and behavior. If Amodei’s piece maps the risks, the Constitution offers an initial view of how Anthropic imagines model-level safeguards.
Henry Kravis & Elad Gil in conversation (video). Interesting discussion on the evolution of PE from KKR’s early days, why Kravis is bearish on new funds but bullish on independent platform companies, and the role of AI in driving returns (important, but not a replacement for excellent human leadership). Not an overly deep examination of the topic, but one that resonates with my own views at present.
Will Manidis: Patient Capital Will Eat The World. Food for thought on the evolution of the VC model, structural changes that have led to the current state of the VC and software markets, and the inevitability of patient, product-agnostic, permanent capital vehicles with no deployment clocks or “mandates to be busy.” Some conceptual resonance with the Great Convergence post I wrote recently.
Platformer: How a New AI tool fixed my single biggest problem with Spotify. Good palette cleanser after you’ve consumed a bunch of information about the extinction of human labor. Casey Newton writes about Spotify’s new “prompted playlist” feature, which allows you to give Spotify exceedingly esoteric guidance in natural language. I’ve been using it for a week and it’s certainly opened up sonic neighborhoods that I had not visited for many years. Worth a spin.
To be continued.