Things built to be read in order.

Lumen

The operator's e-commerce course

Most e-commerce teaching stops at 'find a winning product and run ads'. This starts where that ends: landed cost to the cent, contribution margin that survives an accountant, and the four gates — Launch, Unit Economics, Compound, Exit — that decide whether you have a business or an expensive hobby. Twenty-six modules, a mechanism layer that derives every rule rather than asserting it, and a 13-week programme that tells you when to quit.

The Brut Program

The minimum effective dose, fully specified

A complete hypertrophy program built on one mechanism: a fibre grows when it is fully recruited and under high mechanical tension at the same instant. Four sessions a week, every exercise, set, rep and RIR prescribed, with the technique cues, warm-ups, progression rules, deloads, substitutions and nutrition. Built from the researcher tier rather than from convention — and it says which prescriptions follow from evidence and which are judgement calls.

DUOX

The maximal program

Brut's opposite number. Six sessions a week, 16 weeks of block periodisation, every muscle three times weekly, and a three-week specialisation block where one lagging muscle gets 20+ sets while everything else holds at maintenance. It opens by arguing you probably should not run it — a maximal program at 60% adherence loses to a minimal one at 95% — and then gives you the whole thing anyway.

Meridian

The SAT course

Every skill the Digital SAT tests, taught from first principles and drilled until it holds: a prequestion before you read, worked examples that fade as you take over, and retrieval practice where every answer — including the three you didn't pick — tells you exactly why it's right or wrong. Ends in a multistage-adaptive practice exam and a certificate earned by finishing, not by a score.

Veridian

Pearson IAL Economics, Business & Maths, taught

The whole taught system — every Economics, Business, and Maths paper, one purchase. The taught-course layer alongside Veridian's existing AI examiner (see steinhoff.group/work/veridian) — full lessons, not just marking. All eight Economics and Business papers are complete: Markets in Action, Macroeconomic Performance and Policy, Business Behaviour, The Global Economy, Marketing and People, Managing Business Activities, Business Decisions and Strategy, and Global Business — 60 lessons. All six Maths units are complete too: Pure Mathematics 1–4, Mechanics 1, and Statistics 1 — 72 lessons. 132 lessons in total, from the basic economic problem through comparative advantage and the Marshall-Lerner condition to partial fractions, vectors, and the sine and cosine rules. Every lesson is sourced against the real spec, mark schemes and examiner reports rather than reused from the original build, with every claim independently checked and dozens of real errors — including fabricated 'verbatim' quotes with false citations, arithmetic and logic mistakes, wrong mark boundaries, and near-verbatim reproductions of real exam questions — caught and corrected along the way. Same first-principles, retrieval-tested standard as Meridian, adapted for how each subject is actually marked — essay level descriptors for Economics and Business, full working for Maths — not just multiple choice. Buy the whole system here, or buy a subject separately below.

Leg Verd Econ P3

Leg Verd Econ P3.

AMZ

The Amazon FBA operator's course

Most Amazon FBA content stops at 'find a winner with Jungle Scout and run PPC'. This is the fee stack to the cent — referral, fulfillment, storage, the 2026 fuel surcharge — the three ranking systems that actually decide who gets found (A10 behavioural, COSMO semantic, Rufus conversational), the honeymoon-period launch math, and the account-health discipline that keeps the business alive at all, built entirely inside Amazon's own 2026 mechanics rather than generic e-commerce advice ported over. Eleven modules, Foundations through Scale & Exit, plus a fifty-expert source library behind every claim. Companion to Lumen (archived below), which covers the general dropshipping/whitelabel model this course doesn't touch. This is the standard-tier build: a faithful, well-structured import of a corpus that already carries real provenance-tier tagging, not yet run through the deeper mine-then-cite verification and adversarial review pass planned for a course-max upgrade later — several figures are dated and self-flagged as policy-fragile (tariff sunset clauses, Amazon's annual January fee re-rate); re-confirm those before relying on them.

WAW

Web Agency Weekend — the cold-outreach local web-design operator's course

Find local businesses with no real website (or one nobody's touched since the last decade), build them a working preview before you ever call, and let the demo do the selling. Seven modules: the funnel math derived from its own stage rates rather than asserted as a headline number, the reciprocity mechanism behind the free demo, a 0-100 lead-scoring model, a cold-call script built from named negotiation and sales-psychology frameworks rather than folk wisdom, getting past a gatekeeper, and the concrete hire-by-hire path from one person dialing to a small team running a recurring-revenue book. Compiled and restructured from a working July 2026 operator playbook for a real cold-outreach business — not a hypothetical case study, and it says so where a number is one operator's own result rather than an industry consensus. Standard-tier build.

IDS

Implement, Deploy, Scale — the original operator's e-commerce course

The predecessor to Lumen: ten modules distilling 50+ e-commerce practitioners — dropshipping, white label, product selection, paid media, marketing, MER/measurement, brand building, store CRO, finance/scaling and exit — plus eleven deep dives and per-product playbooks, closing with a dated July 2026 applied strategy case study built against the post-de-minimis, TikTok-Shop-era market. Built June–July 2026. Superseded by Lumen, which rebuilt this same corpus with verified mid-2026 market intelligence, a mechanism layer that derives rather than asserts, and a 13-week execution programme; kept live rather than deleted because a superseded build is a different thing from the new one, not a worse draft of it. Not reconciled against Lumen's or AMZ's newer numbers — read this as e-commerce operating doctrine as it stood before the 2026 shocks reshaped the model it assumes.

AI Lead Automation

Build and sell AI lead response for Dubai real estate agents

A complete build-and-sell playbook for AI-powered lead response in Dubai real estate. Every lesson is built from the actual steinhoff.systems operation — the real code, client emails, and numbers that close. Learn to build the WhatsApp + voice system that recovers 63% of missed leads and converts 47% to viewings.

Income Playbooks

Eight non-resident-friendly, high-ticket business models — mechanism, economics, and the legal line

Eight remote, high-ticket business models, each researched against one shared template: the root mechanism, real unit economics, 2026 regulatory reality, capital required, realistic timeline to first dollar, whether a non-US-resident can legally run it, a step-by-step launch sequence, and a kill-switch framework for when to stop. Every claim carries the confidence tag it was researched with — [Established] / [Directional] / [Speculative] — rather than smoothing hedged research into stated fact. 1 of 8 modules built so far (Insurance, deliberately ranked worst of the eight and built first anyway, as the hardest test of the course's own honesty standard); real estate creative finance, AI consulting, business brokerage, lead-gen arbitrage, remote closing, executive search, and golden visa consulting are researched and queued.

COVER

Selling life insurance without a US license

Five real paths into life-insurance-adjacent income for someone who isn't a US citizen or resident and almost certainly can't get licensed the normal way — compared with worked unit economics, not IMO recruiting-deck marketing: buy-lead IMO agent, own-funnel licensed producer, an offshore-bond advisory career, compliant lead-gen arbitrage, and a content/SEO affiliate asset. The eligibility gate is treated as a hard precondition, not a footnote, and the compliance module is built directly around one lead-gen operator's real collapse — a multi-billion-dollar exit wound down after a nine-figure TCPA settlement over resold consent, then a separate nine-figure FTC judgment a year later — as the mechanical case for why 'we have a checkbox' isn't compliance. A mechanisms module derives the underlying sales psychology from first principles instead of asserting sales folklore, and a single 30-day execution calendar plus a falsifiable quit doctrine replace the usual scattered per-module timelines. Standard-tier build, not yet generalised from the operator it was researched for to a public reader — draft until that pass happens.

AI Agency

Niched AI implementation, readiness audits, and AI-driven marketing ops

'AI agency' gets sold as one thing and is actually three sharing the same infrastructure: niched-vertical implementation, paid AI-readiness audits, and fractional AI-ops retainers — plus a fourth angle this course treats as inseparable rather than a footnote, AI-driven social and marketing operations, with its own tooling landscape, pricing benchmarks, and AI-generated-content disclosure rules. The core argument, sourced rather than asserted: generic technical execution (wiring an LLM to a CRM) is commoditizing fast — a debunked '95% of AI agencies will be dead by 2026' line gets named and discarded along the way — while domain-context judgment about a specific vertical's real bottleneck commoditizes far more slowly, and that's where the actual margin still lives. Five modules, the same [Established]/[Directional]/[Speculative] confidence-tagging as Income Playbooks and COVER, standalone deep-dive standing to Income Playbooks' queued AI-consulting module the way COVER stands to its Insurance module. Standard-tier barebones build; a course-max pass is planned later.

Ad Agency

The full-service, managed-media agency — Google, Meta, programmatic, and AOR contracts

The traditional, full-service, managed-media agency — $10,000/month-and-up clients, Google/Meta/programmatic budgets, agency-of-record contracts — distinct from the smaller-client SMMA model on capital, sales cycle, and what wins a pitch. Derives the make-or-buy mechanism from first principles, works the real math behind commission-vs-retainer-vs-hybrid pay (including the ANA/K2 Intelligence programmatic-opacity scandal), and separates gross revenue from AGI so the margin, churn, and concentration-risk benchmarks in the operating module actually mean something. Closes with sourced 2026 research most 'start an agency' content skips: the Omnicom-IPG consolidation and holding-company AI-driven layoffs, set against the ANA/4A's own finding that independent agencies retain clients longer than holding companies do in the same year — and what that implies for what's actually defensible to pitch. Five modules, the same [Established]/[Directional]/[Speculative] tagging as Income Playbooks, COVER, and AI Agency. Standard-tier barebones build; unreviewed.

SMMA

The social media marketing agency — retainer economics, niche selection, and the saturation debate

Running paid social ads and organic content management as a retainer service for local businesses — the root mechanism, real 2026 unit economics, and an honest verdict on whether the model is actually oversaturated rather than the usual guru-economy hand-waving in either direction. A dedicated niche-selection module traces a widely-circulated 'specialists convert 3x better' statistic back to an unsourced vendor blog post repeated forward through two more blogs with no primary study anywhere, and contrasts it against a real, named 87-agency survey that supports the direction without the fabricated multiplier — the same debunk-before-repeating discipline this platform's other courses apply. Five modules, [Established]/[Directional]/[Speculative] tagged throughout. Standard-tier barebones build; unreviewed.

University Applications

Holistic admissions, aid, and founder-specific paths — the mechanism, not a chances calculator

How holistic US admissions actually work — the rating-scale mechanism the Harvard-lawsuit discovery record revealed, why Early Decision changes the odds and what it actually commits you to, financial-aid mechanics (need-blind vs. need-aware, what a net-price calculator is really estimating), building a genuine 'spike' instead of a scattered activity list, essay frameworks (montage vs. narrative structure), the real application timeline end to end, how to build and use a school-list tracker, and a dedicated module on non-traditional and founder-specific paths (entrepreneurship-focused programs, alternative admissions routes) for an applicant whose story doesn't fit the standard mold. Generalized from source research that originally covered one specific applicant's situation — every named reference, specific figure, and personal case removed and replaced with the underlying, verifiable mechanism. Two modules, nine lessons, confidence-tagged throughout. Standard-tier barebones build; unreviewed.

Sales

First-principles selling — twenty-two mechanisms, eleven frameworks decoded, one playbook

Twenty-two sales mechanisms driven to their psychological and (where real) neural root, eleven named frameworks (SPIN, Challenger, MEDDPICC, Gap Selling, Sandler, and more) decoded back into the mechanisms that actually drive them rather than taught as independent doctrines, and a full eight-stage deal-cycle playbook. A replication scorecard tags every mechanism ROBUST, CONTINGENT, or CASUALTY — and purges three that didn't survive replication (ego depletion, oxytocin-as-trust, subliminal social priming) rather than repeating them because they're famous. Every claim in the mechanism library carries an inline primary citation; the framework-decoding module is original synthesis but mechanically traceable back to it, not asserted independently. Five modules, fifteen lessons, restructured with real editorial work from a working sales-training corpus — not a straight port. Course-creator tier.

MBB

McKinsey, Bain & BCG recruiting — the mechanism, the funnel, and the honest reality check

What McKinsey, Bain, and BCG each say, in their own words, the case interview and PEI/fit interview actually test — and the real recruiting funnel underneath the marketing: resume/GPA screen, firm-specific online assessment (McKinsey Solve, Bain SOVA, BCG's Casey chatbot), first round, final round, offer. Covers the real case archetypes and why a memorized framework is a named, documented failure mode; the mechanical math-and-structuring skill floor; a realistic prep timeline reconciled against both the '6 months' and '3 weeks' claims you'll see; the free-first-then-paid resource landscape including a skeptical, source-checked look at a named paid coaching program; and a kill-switch module that tells you honestly what to fix first if your profile isn't competitive yet. Every specific acceptance-rate and pass-rate percentage circulating in case-prep marketing was checked and, where untraceable to a primary source, named and flagged rather than repeated — the same [Established]/[Directional]/[Speculative] discipline as this platform's other research-grade courses. Barebones first-pass build; a course-max pass is planned later.

Marketing

The mechanism, economics, and measurement underneath every channel-specific course

Marketing as a discipline, not a channel: why paid attention has a price (ad-auction economics), why positioning and differentiation compound, the real economics of CAC vs. LTV, the 2026 channel landscape and how to choose between channels, the brand-vs-performance tradeoff (including a live, named dispute in marketing research — Byron Sharp vs. the Binet & Field 60/40 rule — that this course doesn't paper over), measurement and attribution fundamentals, and the copywriting frameworks that express a position once you have one. Built as the foundations layer this platform's channel-specific courses (SMMA, Ad Agency, WAW, AMZ, AI Agency) assume rather than re-derive, with an explicit boundary against the Sales course: marketing generates and qualifies demand, sales converts it. Six modules, [Established]/[Directional]/[Speculative] confidence-tagged throughout, researched from scratch. Barebones IDS-style build.

AI Automation

Using AI to automate your own business's operations — not selling automation to clients

The other thing 'AI automation' can mean, distinct from AI Agency's client-services business: an operator automating their own operations with 2026-era orchestration tools (n8n, Make, Zapier) and AI agents — invoicing, scheduling, data entry, tier-1 support — with the real economics behind it. No client, no sales cycle, no information-asymmetry premium to capture: just the arithmetic of an hour of labor versus the marginal cost of a tool, done honestly, including the failure modes and the marketing-tier ROI numbers (a '280-520% ROI' claim among them) this research found and explicitly discarded rather than repeated. Six modules, the same confidence-tagging as every other course on this platform, cross-linked to AI Agency at the fork points where a reader's actual interest turns out to be the other one. Standard-tier barebones build.

QUANT

Order flow trading and market microstructure

Not general quant trading — order flow specifically: reading the order book and the tape, order flow imbalance, absorption, iceberg orders, and the regulatory line between legal order-flow analysis and prosecutable spoofing. Built against a peer-reviewed root mechanism rather than retail 'smart money concepts' folklore (named and disqualified explicitly, alongside unsourced iceberg-detection accuracy claims and prop-firm pass-rate marketing), with Jane Street and Renaissance Technologies named as calibration for the caliber of firm this edge matters to — not a claim that either is replicable solo, and no claim about Renaissance's actual current strategy is presented as fact anywhere it can't be verified. Six modules, ten lessons, [Established]/[Directional]/[Speculative] tagged throughout.

Course Selling

The business of building and selling a course — not how to build one

The business model underneath every course-creator success story: why anyone pays for curated, sequenced, accountability-bearing structure when most of the same information is free, and why the single biggest predictor of outcome is whether you already have an audience — not content quality. Compares Kajabi, Teachable, Podia, Circle, and Skool's own current fee schedules against self-hosted and lightweight alternatives, derives pricing and launch-sequence mechanics from first principles, and gives launch economics the hardest scrutiny in the whole course — naming and discarding an untraceable '1-3% of your list converts' benchmark, a self-reported '$1 billion in sales' launch-formula claim, and a cohort platform's own 'completion' claim that turns out to measure something narrower. Closes with real peer-reviewed completion-rate data (12.6% median across 221 MOOCs), an honest refund and FTC regulatory picture, and a kill-switch framework. Six modules, twelve lessons, confidence-tagged throughout. Standard-tier barebones build.

SaaS

Recurring-revenue economics, the real 2026 build landscape, and an honest survival check

Building and selling a software-as-a-service business, starting from the mechanism most 'start a SaaS' content skips: why recurring revenue is worth more per dollar than one-off revenue, and the actual math connecting churn, LTV, CAC, and payback period. Covers the real 2026 build tradeoff (no-code vs. custom, and a genuinely skeptical two-study look at what AI-assisted coding does and doesn't speed up — one study found a 56% speedup on greenfield work, another found a 19% slowdown on realistic work in an existing codebase), pricing model choice by value metric, and the distribution problem stated plainly rather than oversold. The retention module reads four genuinely-disclosed benchmark sources side by side with their commercial interests named rather than picking whichever number flatters. Fundraising vs. bootstrapping gets a real sourced comparison, and the closing kill-switch framework is backed by an honest audit of SaaS survival-rate claims — every popular 'X% of SaaS startups fail' figure traced and found unsourced, replaced with real establishment-survival data instead. Seven modules, thirteen lessons. Standard-tier barebones build.

Building with Claude

Skills, subagents, and workflows, taught from this site's own real files

How this site's own courses actually get built with Claude Code — Skills, subagents, and multi-agent Workflow scripts — taught from the real files already in this repository rather than invented tutorial examples: SKILL.md frontmatter, the workflow scripts under knowledge/workflows/, and two published articles on what broke. Seven modules: how a Skill's description works as a trigger, the agent() primitive and what isolation costs, the pipeline/parallel/adversarial-verify Workflow pattern, a real course build with a real production incident, and an honest-limits module built on live evidence — a production course whose own version history runs past two dozen separate fix passes weeks after a review council had already run, a silent failure that reported a confident wrong answer, and the same-model-reviewer bias adversarial verification doesn't automatically remove. A working build, not a finished flagship — course-max-style depth is a later pass.

Consultancy

The mechanism, pricing, and positioning underneath every independent consulting business — before you pick a niche

The business model underneath this platform's niche-specific consulting courses — AI Agency, Ad Agency, SMMA, MBB — generalized rather than repeated: why a client pays an outside consultant a premium at all, formalized through real credence-goods economics; how hourly, project, retainer, and value-based pricing actually differ on who bears execution risk; and the real mechanism behind why value-based pricing commands a premium, including two widely-repeated 'Deloitte study' pricing statistics checked directly against Deloitte's own site and discarded as untraceable. A positioning module derives why 'I consult on business strategy' fails from the same verification mechanism rather than asserting it as marketing wisdom; a first-client module weighs weak-ties research and a large but methodologically thin survey against this platform's own AI Agency case study rather than repeating either as settled fact; and a structure module works the real leverage economics behind when solo becomes a firm, closing with a falsifiable kill-switch framework built on runway, pipeline trend, and utilization rather than an invented industry benchmark. Explicitly routes readers who already know their niche to this platform's domain-specific courses instead. Seven modules, eleven lessons.

Enterprise Software

B2B, high-contract-value, long-cycle software — and enterprise AI as its own angle within it

The larger-contract-value counterpart to AI Agency, the same relationship Ad Agency has to SMMA: building and selling software into organizations where the buyer is a committee, the contract runs six to seven figures over multiple years, and the sales cycle is measured in quarters, not weeks. Derives the root mechanism from transaction cost economics — switching costs and procurement overhead, not just budget size, set the price — then works the real buying-committee, RFP, and SOC 2 compliance-review mechanics a self-serve SaaS sale never has to survive, and the honest, sourced case for why custom development is usually the wrong answer. A dedicated Enterprise AI module treats the most-cited 2025 AI-pilot-failure statistic with real tiering instead of either repeating or dismissing it — naming the study, its disclosed methodology, and its own named academic critics side by side. States plainly, with real dollar and headcount figures, that this is the most capital- and team-intensive business model on this platform, and closes with a falsifiable kill-switch framework. Seven modules, thirteen lessons.

Research Literacy

Weighing evidence and sources, and reading a paper past its abstract

The two-axis filter this site's own /protocol research dossiers run on in real code — evidence tier (how good the underlying data is) and source tier (who's telling you, and what they stand to gain) — tracked as two separate labels, because merging them into one trust score is the single most common failure in how people evaluate what they read. Works through all six evidence tags and all six source tiers against real, named cases (a whale-longevity paper's CIRBP finding, the taurine reversal, POM Wonderful's FTC case, Coca-Cola's Global Energy Balance Network), teaches where an actual paper hides its own numbers using the power-posing study and its own lead author's public retraction as the worked example, names four checkable red flags (publication bias, conflict of interest, an unreplicated small finding, correlation mistaken for causation) against real cases for each, and closes by running the discipline on one real claim end to end — the marshmallow-test delay-of-gratification finding, from its thin original sample through the replication that shrank it — against what AQA's own EPQ mark scheme actually rewards. Seven modules, seven lessons, evidence- and source-tiered throughout.

PE, IB & VC

Recruiting into private equity, investment banking, and venture capital — and how each business actually makes money

IB, PE, and VC get talked about as one blur when they're three different businesses sharing one seam — banking is the training ground private equity overwhelmingly recruits from, and almost nothing else. Covers investment banking as the best-documented, most standardized funnel of the three (the summer-to-full-time pipeline, real technical/behavioral interview content, and a real 2021 hours survey sourced honestly rather than romanticized or exaggerated); private equity's on-cycle recruiting gauntlet (the paper-LBO-to-modeling-test sequence, and the management-fee-plus-carry economics that explain why a PE firm optimizes for different outcomes than a hedge fund); and venture capital treated as a genuinely different business rather than 'PE but earlier' — fund economics and the power-law return distribution, why VC hiring has no on-cycle process at all, and a dedicated module on Y Combinator sourced directly from its own current deal terms and application mechanics rather than secondhand blog summaries. Closes with an honest three-way comparison of entry difficulty, compensation, and lifestyle. Seven modules, thirteen lessons — a widely-repeated '85% of analysts quit within two years' claim traced to an unsourced blog post and named rather than repeated.

Marketing Agency / Consulting

Two business models, one buyer decision: build it for them, or advise them on it

The gap left after SMMA and Ad Agency: a general, broader-service-mix marketing agency for SMB/mid-market clients, and marketing consulting — most often fractional CMO work — as a genuinely different, advisory-only business. Real 2026 rates for both, the named failure mode where advisory work quietly becomes unpaid execution, and a direct decision guide back to SMMA, Ad Agency, Marketing, and Consultancy. Two fabricated statistics (a claimed McKinsey cost-savings figure, a claimed HBR fractional-CMO growth figure) were checked directly against both organizations' own sites, found untraceable, and named rather than repeated. Six modules, thirteen lessons.