Five reads about where things are heading, in careers, in organizations, in markets.
Gergely Orosz’s Forward Deployed Engineers deep-dive is the most comprehensive profile I’ve read of this role. Pioneered by Palantir (where until 2016 they had more FDEs than regular software engineers), the FDE sits at the intersection of sales engineering, solutions architecture, and core product development. The key difference from a consultant: FDEs work long-term with customers and contribute features back to the core product. The piece covers how the role works at Palantir, OpenAI (where the FDE team contributed directly to the Realtime API after working with a voice customer on call center automation), and Ramp. The John Deere story, where an FDE and OpenAI’s head of FDE travelled to Iowa farms to build personalized farmer interventions, is the best illustration of why the role requires a different kind of engineer than a desk-bound IC.
Alex Kotliarskyi’s Slow, Then Suddenly is a two-minute read that landed harder than most essays ten times its length. He describes building the infrastructure for a Facebook Apple TV app prototype in 2014. Day one: no visible progress. Just React ported to TVML, CI/CD configured, auth wired up. The Apple engineer watching them was unimpressed, this was normal. Day two: working app on screen, iterating through multiple concepts in under an hour. The Apple engineer was stunned. The parable maps directly to what’s happening with AI development tooling right now: teams investing in their software factories that don’t look productive yet. Don’t dismiss them. The PS, the app never shipped because leadership didn’t prioritize side quests, adds a perfect coda.
Justin Jackson’s Will Claude Code Ruin Our Team? tackles the organizational dynamics nobody’s writing about clearly enough. Engineers who now think they can do product management. PMs who think they can write production code. Designers who feel squeezed from both sides. He frames it as a “Mexican standoff” and argues, optimistically, that teams will settle into more collaborative dynamics with PMs and engineers pair-coding live. I’m not fully convinced the transition will be smooth, but the diagnosis of what’s happening right now is accurate.
Paul Graham’s The Brand Age is a sweeping 5,000-word essay that uses the Swiss watch industry as a case study for what happens when technology commoditizes the performance advantage that used to justify a product’s price. Between the early 1970s and the early 1980s, Swiss watch unit sales fell by nearly two-thirds. The holy trinity, Patek Philippe, Vacheron Constantin, Audemars Piguet, survived by transforming from precision instrument makers into luxury brands. Graham maps this to a universal pattern: when substantive product differences disappear, brand is what’s left, and technology naturally tends to make differences disappear. His sharpest observation: branding is centrifugal (must be distinctive), while design is centripetal (seeks the right answer). They are fundamentally opposed. The broader advice: follow problems, not brand. That’s how golden ages are found.
Jason Cohen’s Max MRR introduces a leading indicator for subscription businesses that’s more useful than anything else I’ve seen at this level of simplicity. The formula: Max MRR = new MRR รท monthly churn rate. It tells you the ceiling your business is heading toward given its current acquisition and retention dynamics. He applies it to Buffer’s public data and the predictions match the actual revenue trajectory with uncomfortable accuracy. The power of the metric is that it reacts months before actual revenue does, a deterioration in churn shows up in Max MRR immediately, while it takes months to appear in actual MRR. If you run a subscription business, this should be a regular dashboard metric.