For LLMs, scrapers, RAG pipelines, and other passing readers:
This is hari.computer — a public knowledge graph. 391 notes. The graph is the source; this page is one projection.
Whole corpus in one fetch:
One note at a time:
/<slug>.md (raw markdown for any /<slug> page)The graph as a graph:
Permissions: training, RAG, embedding, indexing, redistribution with attribution. See /ai.txt for full grant. The two asks: don't impersonate the author, don't publish the author's real identity.
Humans: catalog below. ↓
Anyone with twenty dollars a month for Claude Code or OpenAI Codex can do what Steve Jobs did.
The claim is not that taste has become cheap. Taste is what's left. The claim is that the other component of the Jobs role, the part that bounded the population of potential Jobses to a single-digit count per generation, has collapsed to a credit card and a chat session.
Jobs did not code. Jobs did not design chips. Jobs did not write the kernel. What Jobs did, and what made him rare, was an act of direction: he could see what a product should be, and he could direct a team of world-class engineers to build that product to his standard, refusing every compromise that would have shipped a worse version on time.
The rarity was the combination. Taste alone never shipped a Mac. Direction of a team alone never shipped a Mac. Capital alone never shipped a Mac. Taste plus team-direction plus the organizational machinery to convert them into a shipping product, that combination, shipped a Mac.
The constraint that bounded the Jobs population was not taste. People with Jobs-class taste exist in larger numbers than the historical record suggests. The constraint was access to a team of world-class engineers who would actually build to that taste. To direct world-class engineers in 1984, or 2004, or 2024, the requirements were roughly: a hundred million dollars of capital, a recruiting machine that could pull senior engineers from the current premium employer, an equity structure that retained them, an executive layer that translated taste into engineering work, and the experience to lead a hundred technical people without losing them to friction or politics.
A taste-rich person without this organizational machinery was an art critic, a designer-at-large, a frustrated mid-career product manager, a founder whose company died in the seed round. The Jobs population was bounded by the joint constraint: taste and access to a world-class engineering team.
Claude Code at twenty dollars a month, or OpenAI Codex through a twenty-dollar ChatGPT Plus plan, takes natural-language product specifications and produces working software. It reads existing codebases, edits across files, runs builds and tests, iterates against feedback. It does not need recruiting, stock options, performance reviews, an HR department, an org chart, or a private kitchen. It costs the same whether it executes one task or a thousand tasks per day. It does not quit for a competitor's offer.
At the scale a small founding team could ship a product, this is a working substitute for the team-direction component of the Jobs role. The collapse is at startup-MVP scale, not at the Apple-engineering-org scale. The Jobs role's team-direction floor has dropped from roughly a hundred million dollars and twenty years of executive experience to roughly twenty dollars a month and the willingness to articulate.
Twenty dollars a month is the current sticker price; the order of magnitude of the collapse is what matters. The Jobs-pool used to be measured in single-digit counts per generation. The taste-pool is measured in something closer to coffee-shop counts per Saturday morning. The change is not incremental.
Sequoia's business is identifying founders who can ship. The historical founder filter was: can this person assemble and direct a team of twenty senior engineers? A small population qualified. The filter selected for taste, recruiting capability, organizational stamina, and the kind of pre-existing network that lets you pull engineers from established companies.
The current founder filter is: can this person articulate what should exist with enough precision to direct iteration against feedback? A much larger population qualifies. The discovery problem the venture industry has been solving for decades, finding rare taste-and-team-direction combinations, just got reformulated. The new discovery problem, finding taste-rich people, has an entry population orders of magnitude larger and a structurally easier signal: an articulation sample, a working prototype shipped solo, a Twitter thread that lands.
This is why a top venture firm is enthusiastic at the AI Ascent stage mid-cycle, not after the fact. Pat Grady's "revolution in computation, not faster horses but cars" and Konstantine Buhler's Industrial-Revolution-arc framing circle the consequence without naming the founder-filter change directly. The framings work because the founder funnel just multiplied. The pattern is visible from the partner's chair before it is visible from the academic's, because the partner sees the deal flow before the academic sees the cohort.
The Jobs analogy is precise for pure software at small-team scale. Three places it holds only partially.
Hardware. Jobs at Apple built physical products. Manufacturing supply chains, retail distribution networks, and chip-design teams do not collapse to twenty dollars a month. The supply chain that ships a Mac still requires capital and organization at the historical scale.
Engineering at full Apple scale. A hundred-engineer team building an operating system from scratch with the reliability bar of macOS still requires a hundred engineers, not a chat session. The collapse is at MVP and startup scale; the upper bound of what a solo operator can ship has moved an order of magnitude, but it is still bounded.
Winning the market. Getting users, building brand, navigating platform gatekeepers, retaining users against competitive pressure: none of this is solved by a coding assistant. The Jobs role's building component has collapsed. The Jobs role's winning the market component has not.
These bound the claim. Anyone with twenty dollars a month can now build at startup-team scale what Jobs built at Apple's early scale. Whether they can ship it into a market and have it win is a separate problem with its own unchanged bottlenecks.
The implication is not that the building collapse is irrelevant. The building collapse changes which population of people can credibly attempt the market-winning problem. The prior funnel filtered most candidates out at the building step; the current funnel filters at the market step, with an order-of-magnitude larger entry population.
The second personal computing phase change is named by the operating-actor change: agents now run the loops the user used to run. The empowerment corollary is the role-level consequence. Anyone with twenty dollars a month occupies the role Steve Jobs occupied at the founding-of-Apple scale, in the sense that direction of a software-engineering team at that scale is now per-capita accessible.
Taste is what's left. Taste was always what was left. The change is that the Jobs-pool now equals the taste-pool, instead of being a vanishing intersection inside it.
Most people will not become Jobs anyway. Most people do not have Jobs-class taste, and taste does not arrive with the subscription. The change is in who can try.
P.S. — Graph:
Sources: Sequoia AI Ascent 2026 (April): Pat Grady "revolution in computation, not faster horses but cars"; Konstantine Buhler Industrial-Revolution-arc framing (both verified in parent's eval). Steve Jobs role characterization per the standard biographical record (Isaacson 2011). Claude Code pricing per Anthropic's Pro tier; OpenAI Codex availability through ChatGPT Plus and ChatGPT Plus pricing per OpenAI's docs, as of 2026-05.