Head of Content / Creator Ops
The face & hands of the brand on IG, TikTok, YouTube.
This isn't a "here's your role" doc. It's a menu. Everything a small business needs someone to own is on it — with honest notes on who thrives in each seat, who gets crushed by it, and how it actually shows up at StudyFix. Browse it. Star a few. Tell me at the end of the week what feels like you.
Sections 2–4 are deep dives you'll only need after you pick a lane — the actual content library, the actual lecture pipeline, the actual offers & apps as they exist in the repo today. Read them like reference, not homework.
Grouped by Leila Hormozi's five pillars — Marketing, Sales, Product, People, Profit. Tap a pillar chip to filter. Every role card names the Hormozi playbook that goes deep on it, so if a role interests you, you know exactly which PDF to open next.
Pick as many as feel true — you're allowed to want more than one thing. The bars at the end tally which pillars kept showing up in your answers.
Ranked by how often each pillar showed up. Start with the top one — but the second and third are worth a proper look too if they're close.
Tap + on any role to add it. Aim for 3 to try for real this week.
The face & hands of the brand on IG, TikTok, YouTube.
Pays Meta & TikTok to put us in front of med students.
Owns how StudyFix sounds, looks, and feels everywhere.
Gets StudyFix ranking for "how to revise cardiology finals" and similar queries.
Runs the WhatsApp cohorts, DM inbox, and eventually a Skool/Discord.
Hops on 30-min calls with warm leads and closes them into premium tutoring.
Triage every inbound DM, book qualified ones onto Youhana's calendar.
Save calls. Win-back sequences. Off-boarding audits.
Turns lectures into StudyFix decks & MCQs. The core StudyFix IP.
The doing-hands. Editor, videographer, thumbnail designer.
Owns the flashcard app + SBA/OSCE app + lead magnet roadmap.
Makes sure students actually pass. Weekly check-ins, results tracking.
Youhana's second brain. Meeting notes, follow-through, calendar, agenda enforcement.
Recruits, onboards, and holds the culture bar as we hire our first 3-5 people.
Keeps business revenue & personal savings walled off. Reconciles Stripe ↔ Wise weekly.
Runs the price of every StudyFix product. Tests, raises, packages, bonuses.
Builds the dashboards that tell you if you had a good month.
"Where you are not strong, hire the experience. Where you are strong, bring people in that you can train to be your successors. The reality is you want to mix the two, and it is truly dependent on your and your co-founders' skillsets — what do you bring to the organization?"
— Leila Hormozi, 62-Pages of Unfiltered Advice, Q26
If you pick a Marketing, Sales, or Community lane, you're not writing from scratch. The 1_marketing/ directory already has a 30-day content calendar, a hook database, ad hooks bank, email nurture sequences, founder story, objection handling, and cohort welcome packet — Hormozi-audited. Your first job is to know what exists, not invent more of it.
TWO_FOUNDING_OFFERS_DISAMBIGUATION.md.| File | What it is | Use it when |
|---|---|---|
| CaCONTENT_CALENDAR_30DAY.md | 30 days of scheduled posts across channels. Hormozi-audited. | Every morning to know what ships today. |
| HkHook_Database.md | The bank of proven hooks — first 6 words per post. | Before writing any new short-form. |
| AdAD_HOOKS_BANK.md | Paid-ad specific hooks (higher tension, harder CTAs). | Only when paid runs — deferred until organic proves. |
| EmEMAIL_NURTURE_SEQUENCES.md | The full email sequence from opt-in → founding pack sale. | Anyone new lands in ConvertKit → check the sequence is firing. |
| FoFOUNDER_STORY.md | The canonical Youhana + StudyFix origin. Voice reference. | Any bio blurb, About page, podcast intro, DM script. |
| ObOBJECTION_HANDLING.md | Every DM/call objection with the on-brand response. | Every reply to a hesitant lead. |
| CwCOHORT_WELCOME_PACKET.md | Onboarding flow for a new tutor who bought the pack. | Every new tutor within 24h of purchase. |
| NwNEWSLETTER_SYSTEM.md | The full newsletter cadence + templates. | Weekly issue day (currently paused pre-finals). |
| CpCOMMUNITY_POSTS.md | Posts scripted for medical Facebook groups, Reddit, Discord. | Any inbound-community outreach. |
| CfCONTENT_FIRST_DOCTRINE.md | The strategic rationale for content-before-ads. | When anyone asks "why aren't we paying for ads yet?" |
COMPETITIVE_POSITIONING.md)"AnKing is the dictionary. Your lectures are the novel. You need both — but your professor writes the block exam, not the AnKing team."
"Quizlet works from pasted text and a general knowledge base. It can't process a 60-slide PDF, can't guarantee coverage, and doesn't produce ILO or physio cards. Vibe tool, not a clinical tool."
"One-shot prompts don't run a coverage audit. They don't type content by section (SPLIT vs CHUNK). They don't produce four separate output formats. They don't cite source slides. You can make one card with ChatGPT — you can't make 200 without skipping facts."
"You're not wrong that manual cards are high quality. The question is whether 300 hours a year making them is the best use of your prep time."
"Every card in your deck traces back to a specific slide in your lecture PDF. Your student can verify any fact in one click. We can't hallucinate without leaving a citation that exposes it."
— The one-line thesis. Say this. Believe this. Never dilute it.This is the actual StudyFix product. Not a paste-into-Claude workflow — a two-stage Python + FastAPI + Make.com pipeline that turns a lecture PDF into four separate outputs, with a 100% coverage contract enforced by code. If you pick Head of Curriculum or Head of Product, this is your terrain.
"Every card in your deck traces back to a specific slide. If coverage drops below 100%, the system regenerates until it hits it. You get a complete deck or you get nothing."
— FromCOMPETITIVE_POSITIONING.md. This isn't marketing — it's how the Python validator actually works.
Atomic facts, one per card. Recipe A dark-mode section boxes, CHUNK for lists (one cloze per list), SPLIT for interleaved distinct facts. Every card carries a NoteRef pointing to its lecture slide.
Bigger comprehensive cards that consolidate thematically-related sub-topics. No Part 1/Part 2 splits — one card, multiple Recipe A subsections.
Mechanism cards — the "why does this happen" layer. Chain-of-reasoning clozes, expanded scope not fragment recall.
A single readable clinical note per lecture. Ships alongside the .apkg. The read-first companion to the drilled cards.
Access-gate. If they're not on the sheet, Make replies "not enrolled" and the pipeline never fires. This is the reason we can offer the founding pack with a hard 10-tutor cap.
POST /extract → FastAPI 200 immediately → background work›Prompt 1 runs on the cached PDF and produces a hierarchical keyword skeleton with slide-verbatim quotes, testable-unit bullets, and signal markers ([NAMED DISEASE], [PHYSIO CHAIN: N steps], [VISUAL: image type], [DISC]/[CONST] on symptoms).
Then Python's assign_leaf_ids() stamps [L001], [L002]… on every Level 2/3 bullet. The AI never assigns leaf IDs. Python always does. The leaf manifest is the coverage contract.
Python loops every slide. LOW-thinking call: "list every testable fact on this slide, one per line." Missing facts → targeted LOW recovery call: "generate a new skeleton bullet for this fact in the exact format." Up to 3 attempts per slide. Logged failures proceed, don't block delivery.
The student can fix anything wrong in the skeleton before cards are generated. This is where StudyFix earns its "school-specific" claim — the tutor's expertise shapes the skeleton before it becomes 200 cards.
POST /generate → Prompt 2A granular cards (MEDIUM thinking)›Python re-uploads the PDF, creates a fresh context cache (caches are job-scoped). Prompt 2A at MEDIUM thinking produces granular cards as JSON. Python validates: every leaf ID in the manifest must appear in at least one card's NoteRef. Missing leaves → targeted LOW recovery. Up to 3 attempts per leaf.
Every card scored 1–10 on factual accuracy · cloze quality · clinical relevance · anchoring strength · evidence support. Every flag must cite exact slide number + quoted text. Python verifies flags by searching PDF text — unverified flags are dropped. Cards scoring <7 on any axis with a verified flag get one regeneration at MEDIUM. No infinite loops.
Same skeleton + same cached PDF + 2A summary as context. Ship one merged .apkg + clinical_summary.md back through Make S4 → student's inbox.
Every deck passes a 5-check audit before it enters a student's Anki. From feedback_cloze_strict_invariants:
{{c1:: has its }}.}} with no matching open crashes the import.c1..cN monotonic in document order after any SPLIT interleaving.{{cN::…}}.| Skill | What it does | When to fire |
|---|---|---|
| Vdverify-deck | 12-category audit vs eval_cards.md rubric. Detects LGW/PATH/STANDARD lecture mode from PDF + deck header-boxes BEFORE patching. Emits additive patches, drops, and rewrite lists. | Before studying any freshly generated deck. |
| Didisease-index | Multi-lecture deck → single disease-organised revision sheet with C/P · Inves · Mng · Special + a Criteria & Scores tab. Has a mandatory hallucination-audit phase — medical bleed-through is the dominant failure mode. | Once per module for finals triage. |
| Bcblurt-cues | PDF → single dark-mode blurt HTML card with tiered reveal. Vision-OCR every slide, hide everything behind a 60-second soft timer so student blurts on paper first. | Ad-hoc, per lecture, when the student wants active-recall on paper. |
| Dudeck-upgrade | Rewrites an existing deck (any format) into StudyFix v4.0 Recipe A grammar. Ignores input schema — output is a fresh v4.0 TSV. | When a tutor hands over their old deck + the lecture PDF and wants the format modernised. |
| App | What / status | Stack |
|---|---|---|
Lp3_flashcard_app/lecture-pipeline-app/ | PRODUCTION / FROZEN until post-exam (~2026-07-21). The primary pipeline. Single-tenant local, no backend, concierge-only. | React + Vite + Gemini 2.5 Pro |
Fa3_flashcard_app/first-aid-app/ | SHIPPED. Topic-based FA card generator. Different use case from full lecture decks — works topic-by-topic. | React + Vite + Gemini Flash |
V53_flashcard_app/v5/ | BUILT · UNTESTED. Collapses 70-turn verify-deck audit into 3 API calls. Target: ~$1.10/lecture (vs ~$14.30 manual). | Python (chunker + compiler) + Gemini Pro + Opus 4.7 |
Mp3_flashcard_app/MASTER_pipeline_technical.md | The technical bible. FastAPI endpoints, prompt inventory, Make.com scenario wiring, callback payloads, environment vars, Render deploy, troubleshooting. | Reference |
prompts/active/, prompts/archive/, and first-aid-app/src/lib/prompts/ are locked. Never open them unless Youhana says "work on the pipeline prompt." Everything else in the pipeline is fair game — the prompts themselves are load-bearing.
Two founding offers, five apps at different ship-states, one competitor breathing down our neck. If you pick a Sales, Product-Manager, or Profit lane, this section is your ground truth.
First lecture free → $500 for 100 PDFs → $40/mo continuity after that → founding rate locked forever.
Full file: FOUNDING_TUTOR_OFFER.md · GRAND_SLAM_OFFER.md
A separate launch mechanism — different sequence, different close. Do not conflate with the Tutor Pack in any DM, email, or landing page.
CONCIERGE_DAY_FOUNDER_FOLLOWUPS.md.FOUNDING_CONCIERGE_DAY_LAUNCH.md.TWO_FOUNDING_OFFERS_DISAMBIGUATION.md.Read the disambiguation memo before writing any offer copy.
| App | What / current state | Status |
|---|---|---|
| Lmstudyfix-lead-magnet | React + Vite + TS + Vercel + Supabase. Has api/anki, api/bank, api/email, api/cron, api/auth, api/account endpoints. 92/92 tests passing. Public ship deferred pending AMBOSS/UWorld design pivot (active since 2026-06-19). Prod currently on v1 prompt. | Shipped-but-held |
| Lplecture-pipeline-app | The primary flashcard pipeline. PDF → 4 outputs. Single-tenant, local, concierge-only. Rebuilt 2026-05-20. | Frozen · post-exam |
| Fafirst-aid-app | Topic-based FA card generator. Different use case — topic-by-topic, not slide-by-slide. | Shipped |
| V5v5 collapsed-audit pipeline | 3-call audit pipeline replacing the 70-turn verify-deck. Cost target ~$1.10/lecture (vs ~$14.30 manual). | Built · untested |
| MsDisease-index micro-sites | Per-module revision sites at Desktop\Sites\<topic>\, deploy at thestudyfix.com/<topic>/. Mirror the induction-site convention. |
Ongoing |
GEMINI_TEST_KEY has been 429ing since 2026-06-19. Blocks smoke tests. Awaiting top-up.panicbank.com — solo KCL/Newgiza founder. Same demographic, similar angle. Drives every "ship before they do" call we make.
Drop one lecture PDF, get the full output back at zero cost. Trust-builder — if the cards aren't better than manual, they walk.
100 lecture conversions — a full academic year of one subject. Founding rate locked forever. First 10 tutors only.
After the pack. $4/lecture continuity rate. Cancel anytime. The compounding revenue line.
"AnKing was made for boards by strangers. Your professor writes your block exam. Use your lecture." Nobody else makes this claim. It's unoccupied territory — every creator either defends AnKing or attacks Anki as a concept. Own it.
—COMPETITIVE_POSITIONING.md