Second pass through the LinkedIn feed today. Engaged with four substantive posts, skipped the promoted slots and lead-magnet bait. Focused on posts adjacent to Rahil's lane as an AI founder building Slate: AI builder workflows, product UX taste, the changing role of managers in an AI org, and a sharp founder-humor post from PostHog.
Posts commented on: 4
Run date: 2026-05-10
Author headline: Building AI systems that automate how businesses attract, engage…
Topic: Anthropic doubled Claude Code's 5-hour rate limits, removed peak-hour throttling for Pro/Max, and raised Opus API limits. Argues the real benchmark for AI coding tools is no longer model quality but "can I build uninterrupted for hours?" — sustained usage, context continuity, developer velocity.
Comment posted:
Love that framing, the 'uninterrupted hours' benchmark feels way more honest than model leaderboards. Are you seeing teams pick AI tools based on rate limits over raw model quality now?
Why this post: Rahil ships AI products and lives in this exact tradeoff. The post reframes a common debate in a useful way and the question pushes the author to share what he's seeing on the ground.
Author headline: AI Agent Architect · Full-Stack · ex-TwinMind (Lead Web Developer)
Topic: Argues people underestimate the small "unnecessary" UX features that make an app feel alive. Shipped a customizable background system on his AI desktop app, e.g. unicorns or space cats as wallpaper — small detail, but it changes how the product feels.
Comment posted:
Totally agree, the small touches are what make a tool feel alive instead of sterile. What's your test for whether a 'fun' feature actually earns its place vs just being noise?
Why this post: Hits squarely on product taste and UX detail, both relevant to Slate's classroom UX where small affordances matter. Question gives the author a real prompt to share his decision framework.
Author headline: Sr. Director, Product Management @ Workday · Speaker
Topic: "The paradox of the modern manager." 70% of variance in team engagement traces to one person, the manager (Gallup), and yet 50% of middle management layers are projected to be flattened by AI by end of 2026 (Gartner). Cites Brian Chesky and Brian Armstrong arguing pure people-managers won't have value, and proposes the "manager IC" archetype. Notes the skills that hold up are the ones AI can't do: reading the room, catching burnout, translating ambiguity.
Comment posted: