Projects

An opinionated map of elite institutions, sector prestige benchmarks, and the eight real mechanisms by which people build significant wealth — plus what actually moves the needle on a CV, argued honestly as opinion rather than dressed up as a ranking.

An exhaustive sector-by-sector directory of the firms that read as prestigious in finance, law, tech, luxury, media, industrials, and consumer/healthcare — the full companion reference to the pathways piece above.

Five real paths into the life-insurance money supply, compared on the same terms: what actually blocks the buyer, what blocks you, and how many weeks stand between "start today" and the first dollar landing.

About four underlying ideas produce every SAT grammar rule; this derives the rules from those four instead of handing over a pile of disconnected ones to memorize, then drills them with 32 digital-SAT-format questions that force naming the rule before revealing the answer.

1,694 real, verbatim College Board Digital SAT Question Bank items across all 30 officially tested skill points, with a worked explanation on every choice — extracted directly from the commits that ported them into Meridian, not reconstructed from memory. College Board's own content; not affiliated with or endorsed by them.

A second rubric for a different question — not "what earns fastest" but "what actually scales into a company" — comparing a productized service, a cold-start SaaS, and a service-to-SaaS hybrid on ceiling, capital efficiency, moat, and fundability. Service → SaaS wins: start on services for cash and credibility, then productize the part that repeats once the pattern is proven.

Twenty-three ways to make money — tutoring, reselling, video editing, AI automation, Amazon FBA, OnlyFans management, and eighteen more — scored on a shared weighted rubric (startup cost, speed to first income, realistic income, competitive room, downside risk, Wharton application value). Tutoring and flipping/reselling win on a near-zero-capital, time-rich profile; paid-ads dropshipping and Shopify DTC score worst on capital efficiency; OFM is disqualified outright and finishes last.

The reference document that set the bar for how Meridian's own lessons should teach: every tested grammar rule derived from three underlying signalling mechanisms, never stated as a fact to memorize. Kept as the worked example behind this site's "derive, don't assert" content rule.

A pipeline that turns a person's own output into a structured, citable file you can put a question to — every belief and rule pointing at a video and a timestamp — and then measures it against held-out material instead of asserting it works. Six subjects extracted, 4.25M words on the largest, and zero converted so far: the machinery that scores a persona got finished before a single persona went through it.

Every multi-agent workflow actually run building Meridian, Veridian, and the rest of this project — not templates, the real scripts, with the failures and fixes still visible in the prompts themselves. Lives in the codebase's own knowledge base rather than only in a session transcript that eventually ages out.

The exhaustive sibling to course-builder — 27 reusable prompts, a reconstructed quality rubric, nine course-building-specific orchestration patterns, and a full competitor/UX research base, for a build meant to be the best in its category rather than just shippable.

The distilled process behind the workflow archive above: when a workflow beats a single agent, why research and implementation are two calls not one, and the exact fix for a silent output-token-ceiling failure that took three attempts to diagnose.

How to build a paid course on this platform's engine without reinventing commerce, auth, or the assessment model it already has — the repeatable version of rebuilding Meridian end to end.

Course-writing rules that trace to actual evidence, not intuition — retrieval practice, spacing, format fidelity, and an explicit do-not-build list for the engagement patterns that measurably make learning worse.

A real, registered venture (Steinhoff Group LLC, working name Ramsden) building AI-assisted enterprise and ERP-adjacent software for other businesses — running in parallel to this site rather than documented on it yet.

Research drafted for a paid-advertising agency model — media buying and campaign management sold as a retainer service — not yet turned into a public course or page.

Research drafted for the classic SMMA model — running paid social campaigns for local businesses. Scored 12th of 23 in the Business Model Evaluation above; not yet built into a public course.

The same tiered-encyclopedia format applied to Zegna — suits & tailoring, knitwear, shirts/polos, and outerwear, ranked S through D by rarity and resale. Multi-colorway pieces are a single product card with a swatch picker rather than one card per colorway. Not affiliated with Zegna; resale figures are estimates.

The same tiered-encyclopedia format applied to Loro Piana — cashmere & wool knitwear, polos/shirts, outerwear, and trousers/suiting, ranked S through D by rarity and resale. Multi-colorway pieces are a single product card with a swatch picker rather than one card per colorway. Not affiliated with Loro Piana; resale figures are estimates.

The Chrome Hearts encyclopedia's format applied to Polo Ralph Lauren — 100 products (143 real colorway variants) across rugby, knitwear, outerwear, polos/oxfords, denim/chinos, and shorts/swimwear, ranked S through D by rarity and resale. Multi-colorway pieces are a single product card with a swatch picker instead of one card per color. Not affiliated with Ralph Lauren; resale figures are estimates.

A tiered reference guide to Chrome Hearts tops — 49 pieces ranked S through D by rarity and resale, each with a verified real photo (and a front/back flip toggle where a back-view shot exists), retail, resale range, release date, and a buy link. Not affiliated with Chrome Hearts; resale figures are estimates.

Two reference documents on elite institutions and wealth-building pathways, turned into two diagrams instead of two more lists: a branching flow chart of the eight real ways people build significant wealth, and a filterable grid of the benchmark firm for forty-plus sectors.

A Claude Code session mapping two from-scratch paths to a first dollar — AI automation freelancing and indie product-building — stage by stage (learn, build proof, first money, get customers, economics), with customer-acquisition tactics for each. Casual session notes, not sourced research — the ai-agency course covers the AI-consulting side in far greater, sourced depth.