We (Matt and Shray) read hundreds of articles on company building, venture investing, and self-management and curate the best ones into a weekly digest to help founders and operators stay on the top of their game.
Better thinking
The third era of AI software development (5 minute read)
This went viral over the weekend from Cursor’s founder. Michael argues we're entering a third era of AI coding. First came tab autocomplete, then interactive agents. Now autonomous cloud agents handle larger tasks independently, returning reviewable artifacts. 35% of Cursor's own PRs already come from these agents. Developers are becoming managers of agent fleets. The thing everyone has been saying would happen is finally starting to happen.
Legible to yourself (7 minute read)
A riff on Will Manidis’ “Legible to capital” essay, arguing that before you can make your company legible to investors, you need to be legible to yourself. The people who thrive have no gap between who they are at work and at home, and everyone around them knows exactly what they’re getting.
Sam Kriss visits San Francisco to profile a new generation of tech founders who’ve been rewarded not for skill or intelligence, but for “agency” (the ability to just do things, relentlessly). He profiles three characters who embody this in very different ways, and the picture that emerges is funny, unsettling, and a little sad. It’s a long read, but well worth it.
Operational tactics
The hard truth about enterprise AI adoption (5 minute read)
Enterprise AI adoption fails routinely because of three common reasons: inadequate focus on the basics of the problem (KPIs, clear ownership, business outcomes), underdeveloped data infrastructure (quality, middleware, access rules, connectors), and poor communication (insufficient clarity on roles, expectations, and incentives). Here’s what you need to do this year: 1. Spend time on straightforward use cases for customers because boring often beats brilliant 2. Develop AI-literacy so you can equip your teams with the judgment, vocab, and mental models needs to work alongside AI effectively 3. Codify the team’s AI supply chain (Data → models → tools → workflow → compliance → feedback loops) so you (and your people, processes, and infrastructure) can ship faster, safer, and cheaper than those still running bespoke builds.
Who owns pricing (3 minute read)
When most founders think about pricing, few ponder about system architecture. But what exists under the pricing hood dictates how fast you can test and how confidently you can scale. And in the AI era, those who treat pricing like a product will win. A sensible place to begin is to think about four modular pricing pillars. Pillar one is the unified product catalog that ensures global changes appear for billing, provisioning, analytics, GTM, etc. Pillar two is all about decoupling pricing logic from code so business users can change packaging seamlessly. The next pillar focuses on real-time metering to give customers transparency and predictability. The final pillar enables non-engineers to manage monetization, allowing them to create new plans/SKUs, adjust pricing tiers, and trigger experiments on the fly. In principle, these pillars dramatically increase agility.
Location, location, location (3 minute read)
The number one rule in real estate is location, location, location. For deeptech startups, this rule applies even more so, and is often dictated by the physics of what the founding team’s building. However, geography doesn’t bind uniformly across a company’s lifecycle. At the early stage, the structural limit is lab access. At later stages, the limitations shift to manufacturing and proximity to customers, regulators, and capital. So founders should focus on what their dominant constraint is for the next 24 months and determine where that constraint is already someone else’s routine problem. This helps to reduce execution risk and improves company survivability. The bottom line in real estate is that the agent knows you’re not buying the house, you’re buying the neighborhood. In deeptech, it’s no different.
Refer and we’ll send you our favorite books as a “thank you” for spreading the word.
Private investing
“We don’t believe in VC value add or platform teams” (2 minute read)
Shots fired. We’re sharing this to be provocative (it gets the people going!). Slow Ventures argues that VCs don’t add value outside of capital. They embrace that directly. And they go one further by saying that platform teams are just smoke and mirrors, designed to make a fund feel good, but not to really be useful to founders. The core of the argument is that, invariably, the platform team spends all the time with the lowest performers.
Infinite games (10 minute read)
Our friend, Mario, the founder of The Generalist, is joining Hummingbird VC as a partner. I’m incredibly excited for his move. He’s not only a wonderful writer, but I think an exceptional investor. Much more thoughtful and measured than many of his peers. He breaks down why he decided to make the move here, and it positions Hummingbird in a very pleasant light.
2026 Tech private markets analysis (4 minute read)
Here’s a report produced by Caplight for people who’re actively thinking about what’s next (job-seekers, people considering entrepreneurial ventures, investors, etc.). Caplight aggregates closed secondary trade data and live buy/sell orders across 1,000+ venture-backed companies. This report draws on Caplight’s proprietary database of ~5,000 closed trades ($10B+ cumulative volume) and 17,000+ order snapshots. Here’s what stood out in the report: application-layer AI companies are your best bet to catch a unicorn, selling before IPO often beats waiting through lockup, and working at a private AI company may be your only way to capture AI upside.
Career management
What marketers are building with Claude Code (30 minute read)
Four marketers share what they’ve actually built with Claude Code and Cowork (mostly internal tooling), from homepage graders to competitive intel scrapers. The piece walks through each build step by step, with practical tips for getting started and a glossary to cut through the jargon.
Anthropic’s Jenny Wen on why the traditional design process is dead, how engineers are forcing the role to evolve, and what designers must do to stay relevant in the age of AI.