For: Alexander "AC" Hampton · From: Atlantic Automation · Date: August 17, 2026
The Consulting Brain is a knowledge base built only from Supreme Ecom's own material — curriculum, case studies, SOPs, and consulting-call teachings — with a search engine and a rulebook on top. When a student asks Claude a question, the brain finds the right Supreme Ecom material, judges honestly whether that material covers the question, and hands Claude the passages plus binding rules on how to use them. Claude writes the answer; the brain decides what it's allowed to say. Everything a student is told traces to a named Supreme Ecom lesson, case study, or process — never to generic internet advice.
Validated on production (August 17): 19 of 19 golden questions pass. Grounded answers on methodology, honest refusals off-topic, failure-aware product validation, and correct answers about the mentorship offer.
Three inputs and two policy calls. Everything else in this document is context for these.
| # | We need from you | Why |
|---|---|---|
| 1 | More student outcomes — transcripts, write-ups, or notes for students other than Jakob, and failed products/stores | Today's evidence is one student's story plus three headline results. "How our most successful students do it" needs a bench. |
| 2 | Screenshots of top student store pages (Julian's product-page patterns) with a one-line caption each | The store-guidance tool is built to show them; it has nothing to show yet. |
| 3 | Approval of what members may see — student first names, revenue figures, and which archive material is current | Case studies carry real names (first + initial) and real numbers. |
| 4 | Ranking policy — should Supreme Ecom's taught positions always outrank student evidence in retrieval, or rank by relevance with taught positions led at the answer layer (V1 today)? | Our recommendation: keep V1 until #1 lands. Either way, it's a setting we flip. |
| 5 | Partial-coverage behavior — when material only partly covers a question, V1 answers from the nearest Supreme Ecom principle with an explicit "I'm generalizing" caveat. Keep, or refuse harder? | Our recommendation: keep — a caveated Supreme Ecom answer beats a refusal for a paying student. |
36 documents, distilled from the 45 raw files (~2 MB) received August 5 and 12: the curriculum export, twelve cohort call decks, call transcripts, and Jakob's full client transcript. All dated 2026.
| Category | Docs | What it holds |
|---|---|---|
| Curriculum — what Supreme Ecom teaches (all marked canonical) | 22 | Product validation criteria · four-pillar system · niche strategy · TikTok-discovers/Meta-monetizes · store structure · offer strategy · branding · five-question ad framework · campaign systems · kill/keep rules · pivot vs optimize · funnel diagnosis · scaling · AOV · profit systems · audience intelligence · operations · system mastery · mindset · the 90-day mentorship offer |
| Case studies — student outcomes | 5 | Jakob ×4 (energy gum ~$100k in ~90 days; first-product stall; subscriptions; five-country testing) · Maurice ×1 (organic TikTok, zero ad budget) |
| SOPs — step-by-step processes | 4 | Product research system · week-1 setup · CBO first launch · store conversion checklist |
| Consulting-call teachings | 4 | Ad budget rules · creative fatigue protocol · pixel preservation · retargeting structure |
| Testimonials | 1 | Published results from supremeecom.com: Ashley F. $5,200/2 mo · Michael K. $1M/9 mo · Alex F. $22k/22 days |
| Screenshots | 0 | System built; awaiting your images (#2) |
The offer, as loaded (canonical, dated August 2026): 90-day mentorship · one cohort call per week · one one-on-one call per week · seven-day-per-week chat support · Blueprint 6.0 included until Blueprint 7.0 is released.
Dating. Every document carries a date. Newer dated material outranks older, and undated material ranks as old — so a freshly uploaded lesson always beats last year's. Anyone loading material without a date is warned at upload time and told where the date goes. Material older than 18 months is presented as "what Supreme Ecom taught at the time."
How the brain finds the right material. Every Supreme Ecom document is split into short passages, and each passage is stored with a "meaning fingerprint" — a numerical signature of what it says, not just the words it uses. When a student asks a question, the question gets the same kind of fingerprint, and the brain pulls the passages whose meaning sits closest to it. That's why "my ads get clicks but no sales" finds the funnel-diagnosis framework even though the student never said "funnel." It then measures how close the best matches are: close enough and corroborated by a second passage means "covered"; a weak match means "we don't teach this" — and the brain says so rather than guessing. Only then does Claude write, under rules that put Supreme Ecom's taught position first and cite the source by name.
In words:
| Tool | What it does | Searches |
|---|---|---|
| Search knowledge | Any business, store, or product question | Everything |
| Store guidance | Supreme Ecom's taught position on layout, collections, pricing, promotion — with student-page screenshots when available | Curriculum, case studies, SOPs |
| Validate a product idea | Walks the idea through the taught criteria one by one, then names the documented failure it resembles and the successes it matches | Three searches: criteria, successes, failures |
| Find case studies | Proof and precedent from students in a comparable spot | Case studies, testimonials |
| Research method | The taught way to do a task — find products, gauge demand, spy on competitors | Curriculum, SOPs, call teachings |
Plus a "check my access" tool for students, and five instructor-only tools to add, update, and remove documents.
V1's design: rank by relevance to the question, with Supreme Ecom's taught positions given precedence at the answer layer. The brain does not force "curriculum beats case study" inside the search — on today's corpus that would bury the Jakob evidence that makes answers concrete. Instead, taught positions are separated out and Claude is bound to lead with them as Supreme Ecom's position. Dated newer material gets a further edge over older or undated material.
Observed mix in live tests (share of passages returned):
| Question type | Supreme Ecom's taught positions | Student evidence | General knowledge |
|---|---|---|---|
| Methodology ("what margin do I need?") | ~80% | ~20% | 0% — not permitted |
| Product validation | ~55% | ~45% | 0% — not permitted |
| Off-topic ("German VAT") | — | — | 0% — refused |
Read these as a snapshot of the corpus as provided, not a fixed setting: as you supply more case studies and testimonials, the evidence share rises. If you'd rather lock in a fixed priority order, that's decision #4 and it's a setting we flip.
The brain scores how closely the material matches the question and applies three thresholds, calibrated against Supreme Ecom's real material (real questions score clearly above the line; adjacent-sounding-but-untaught topics clearly below). Thresholds are settings, re-tuned after each batch of new material.
| Verdict | Rule | What Claude is told |
|---|---|---|
| Covered | Strong match, corroborated by a second passage | Answer from these; name the material; add nothing they don't support. |
| Partly covered | Some match, not corroborated | Answer only what the material supports and say what it doesn't address — or, if no taught position was found, "I'm generalizing from a Supreme Ecom principle." |
| Not covered | Weak match | "The Consulting Brain doesn't cover this." Name what it does cover. Never answer from general knowledge. |
Access. Membership is re-checked every 5 minutes; if the membership system is briefly unreachable, a student verified within the last 24 hours keeps access. Refusals name the reason and the fix.
A. "What profit margin do I need on a product before it is worth testing?" Covered. Led by the canonical validation criteria — minimum 2.5x landed cost, ideal 3–4x; bad margins = invalid product — supported by the scaling, profit-systems, AOV, and product-research frameworks, with Jakob's first-product stall as the cautionary case. Claude leads with the 2.5x rule as Supreme Ecom's position and cites Jakob's case by name.
B. "A trendy fitness product from TikTok — already selling for other dropshippers, looks saturated, but gets views easily." (Validate a product idea) Covered. Criteria from niche strategy, the validation checklist, four pillars, and the TikTok/Meta split; successes matched from Maurice and Jakob; failure matched to Jakob's first product — so Claude is instructed to walk the criteria one at a time and name the failure it resembles.
C. "How do I handle quarterly VAT returns in Germany?" Not covered. Nothing returned; Claude is told to say the brain doesn't cover this, list what it does cover, and not improvise.
D. "How long is the mentorship and what calls do I get each week?" (added today) Covered, canonical. 90 days · one cohort call/week · one one-on-one/week · seven-day chat · Blueprint 6.0 until 7.0.
Atlantic Automation · August 17, 2026