Marketing Strategy & GTM
Visual launch plan — checklist, timeline, per-account & per-platform, bios/links, and the visual system. Phase focus: Instagram + Reels + TikTok.
Profiles & positioning
Bios + highlights + grid themes (from the mockups)
Autoridade de operador — o unico no nicho que de fato fechou $1B+ e transforma pesquisa institucional real em workflows que um analista consegue re-rodar. Dono da faixa 'workflow institucional reproduzivel'.
CEO @ Wall Street Prompt | Investor, $1B+ in deals closed | AI Research Fellow, Harvard | I turn institutional research into AI workflows analysts can reproduce
Free research workflows + the weekly prompt -> tap below · davewang.ai (newsletter landing -> lista de 41k; botoes secundarios: Free Course, X @davewangmia, YouTube @davewang_ai, wallstreetprompt.com)
Grid: AI for Equity Research · 3 Prompts Every Analyst Needs · How Hedge Funds Use AI · Build an AI Finance Workflow · Replace Repetitive Research Tasks…
Handle mockup 48.7k. Funil sexta para a lista de 41k. Keywords proprias (nao mexer): PROMPT, FILING, STACK, TENK, WORKFLOW, TIE-OUT, TOOLS.
Acessibilidade + verificacao — mesma pedigree Goldman/SoftBank e $1.5B deployed que Dave, mas o diferencial e plain-English, no-code, verification-first (NAO senioridade junior).
Ex-Goldman Sachs, ex-SoftBank ($1.5B deployed). I teach AI for finance in plain english. Educational only, never financial advice.
Subscribe + grab the free guides (link below) · linktr.ee/felipesinisterra (newsletter 25k+, biblioteca de 72 guias / '8 Files to Master Claude in 2026', Free Course, LinkedIn, X @fasinisterra, wallstreetprompt.com)
Grid: AI for Finance, Plain English · 5 Prompts for Analysts · How Goldman Analysts Use AI · AI Workflows for Finance · Stop Doing Research Manually…
Handle mockup 28.7k. Canal IG net-new pra ele — converter audiencia fria com a biblioteca de 72 guias como isca de DM. Keywords proprias (zero overlap com Dave): REPS, VERIFY, TUTOR, BLUEPRINT, ANALYST, DCF.
Marca institucional sem rosto — o sistema operacional de pesquisa com IA pra mercados. Em hold ate o estilo proprio ser fechado; founder-led e o motor primeiro.
AI for the people who do the work in markets. Reproducible research workflows, verified prompts, and the standard for using AI in finance. Educational only — human judges, AI builds.
The AI research operating system for markets -> wsprompt.com · wsprompt.com (knowledge base institucional; founders Dave & Felipe linkados)
Grid: The 8 AI Tools I Open Every Day · Reproducible Research Workflows · Verify Before You Trust · The AI Finance Standard · Human Judges. AI Builds.…
Em hold (2-3 posts/semana). Series 'Second-Order' (qui) e save/follow, SEM keyword de DM (preserva o tom institucional). 1 celula institucional de ads (~12% do budget).
Reels — visual direction (locked)
One locked identity per founder — consistency IS the brand
Set: Home-office cinematografico escuro: estante desfocada a esquerda + LED amber quente, arte abstrata escura atras, estatua de touro + globo + vela acesa + livros (Daily Stoic / Tools of Titans), MacBook em mesa de madeira, caneca preta. Shallow DoF (fundo em bokeh). Key suave no rosto, abajur quente a direita, fill frio escuro. Mesmo fundo, mesmos props, mesma luz sempre.
Wardrobe: Vest meio-ziper navy/charcoal com logo teal 'wallstreet prompt>' no peito esquerdo, sobre camisa azul-clara/branca. Zero variacao entre fundadores ou videos. O vest E parte do logo — tem que estar no quadro e o logo legivel no crop da thumbnail.
Framing: Vertical 9:16, crop mid-chest, sujeito centralizado, nivel dos olhos, olhando direto pra camera. Mesmo crop e headroom sempre (rosto cai no mesmo ponto na grade e no player). Sem b-roll que quebre o talking-head centralizado.
Captions: Burned-in palavra por palavra, sincronizadas com a VO. EXATAMENTE 1 keyword por beat em verde WSP #16C784 — nunca dois verdes, nunca outra cor. Peso condensado/pesado, alto contraste, margem segura centralizada pra nao clipar em nenhum device.
First 3s: Hook como texto grande na tela E falado no mesmo instante no 1o segundo — sem aquecimento. Usar shape provado (choque de numero/custo, myth-bust, I-built-this, vs/framework, listicle, comment-keyword). Liderar a linha com a frase de busca US literal ('analyze a stock with AI', 'read a 10-K with ChatGPT') pro hook virar descoberta.
• Mesmo set, mesmos props, mesma luz em todo Reel.
• Vest navy 'wallstreet prompt>' sempre on-frame e legivel na thumbnail.
• 9:16 mid-chest centralizado, nivel dos olhos, mesmo headroom.
• 1 unica keyword verde #16C784 por caption beat — nunca duas, nunca outra cor.
• Hook na tela + falado no 1o segundo, com frase de busca US literal.
• HeyGen avatar identico por fundador — nunca regenerar.
• Rodape compliance: educacional, sem buy/sell/hold.
Weekly posting
4x/week per founder
SEG — Workflow/Ensino (slot de SAVES): passo a passo reproduzivel com beat VERIFY-THIS; termina em 'salva isso pro seu proximo earnings day'.
QUA — Hook/Contrarian/Myth-Bust (slot de ALCANCE): motor de shares; fecha com 'Human judges. AI builds.'
SEX — Comment-Keyword Isca (slot de FUNIL): 'Comenta [KEYWORD] e eu te mando o [entregavel]'; 1 keyword unica testada ponta a ponta antes de postar.
DOM — Prova/'Eu construi isso' (slot de AUTORIDADE): 'AI vs the Analyst' ao vivo, 4 min vs 40, com passo VERIFY; sem promessa de retorno.
Stories: Quase todo dia, ao redor de cada Reel: teaser antes, reforco depois, bastidores do VERIFY, caixa de pergunta semanal (vira material de QUA), link sticker pro lead magnet so com asset vivo, prova social de DMs/saves.
Batch: Filmar os 4 Reels da semana em UMA sessao por fundador (set/vest/framing/avatar travados). Fluxo: (1) escolher 4 referencias PROVADAS da Rip List e clonar hook+duracao; (2) escrever 4 scripts com hook na tela no 1o segundo + 1 keyword verde #16C784; (3) gravar tudo de uma vez; (4) burn captions e agendar nos horarios fixos US-ET.
Ads per post
Boost only what earned organic
Organico primeiro, sempre — nunca impulsionar frio. Cada asset publica organico e precisa GANHAR a promocao no gate de 72h: (1) saves+shares >= 8/1.000 reach; (2) hold-rate aos 50% >= 25% ou completion acima da mediana; (3) >= 1 comentario-keyword real. Passou -> ad nativo 9:16 sem polimento. Falhou -> Content Radar (aprende/itera), nao impulsiona. Objetivo desta fase = SINAL (engagement/video-views), nao conversao. Compliance em toda criativa.
• Publicar organico e medir 72h contra o gate de 3 criterios.
• Passou no gate: boostar como ad nativo 9:16 (Lane A — promover vencedores), espelhamento organico, zero ad-polish.
• Em paralelo: Lane B lab de validacao com <=4-5 celulas ABO, hook-test (3s) primeiro, uma variavel por celula.
• Budget USD 350-500/sem (USD 50-70/dia); semana 1 throttle USD 35-40/dia (~USD 250) por conta nova; subir so com delivery estavel.
• Split 45% Felipe / 35% Dave / 12% WSP / 8% retargeting (ativa ~sem 2-3 quando ha audiencia de video-viewer).
• Piso por celula USD 8-12/dia (>=1.000 plays de 3s por janela); nunca editar ad set no meio do learning.
• Falhou no gate -> Content Radar como nota de aprendizado, sem gastar budget.
Prioridade maxima = HOOK TEST de 3s: mesma base de Reel, variando so a abertura (formula/linha de hook), uma variavel por celula, pra comprar leitura limpa de hook-rate e retencao antes do asset ganhar espaco no calendario.
Hooks & covers
Thumbnails + hooks
Hook falado + texto grande na tela no 1o segundo, liderando com a frase de busca US literal pra virar descoberta. Sem 'hoje vou falar sobre'.
• Copy-This-Prompt — 'Esse prompt transforma [ferramenta] no seu [papel]'; verde ASSISTANT; CTA comenta PROMPT.
• Autoridade + Compressao de Tempo — 'Eu [credencial]. Prompt que uso pra [tarefa] em [N min] em vez de [10N]'; verde 4 MINUTES; so numeros reais.
• Contrarian 'O Erro' — 'O erro que [audiencia] comete com IA: [comportamento errado] antes de [skill]'; verde VERIFY; CTA comenta VERIFY.
• 'If I Started Over' — 'Se eu reconstruisse [edge] do zero em 2026 so com IA, esse e o roteiro'; verde BLUEPRINT.
• Comment-to-DM — 'Comenta [KEYWORD] e eu te mando o prompt que [constroi artefato]'; ex DCF; rotar ANALYST/PROMPT/DESK/REPS.
• Listicle Numerado + Save — 'As [N] ferramentas/skills de IA que [abro todo dia / firmas contratam]. Salva isso'; verde = o numero.
• Second-Order / Pattern-Interrupt — pro viewer sofisticado WSP-brand.
Creator learnings
From the analyzed creators
• maestroprompts (115L/71C): 'copy-this-prompt' entrega asset usavel = saves + DMs altos -> Formula 1.
• Felipe $1.5B (hook 154): credencial real torna o claim impossivel crivel + contraste 4-vs-40 para o scroll -> Formula 2, so numeros verificados.
• the.rachelwoods (save-driven): nomear o comportamento errado do viewer e reframar verificacao (nao previsao) como o edge -> Formula 3 (diferencial do Felipe).
• sabrina_ramonov ('If I started over' / '8 tools'): restart-frame transfere credibilidade de operador; contagem especifica + 'save this' = top save rate -> Formulas 4 e 6.
• ishaan_576 + nick_saraev (771L/514C): artefato nomeado (DCF/FILING/TENK) faz a keyword valer o esforco e roteia direto pro ManyChat -> Formula 5, template mais proximo do WSP.
• jarvisinvest (validado): second-order / pattern-interrupt pro viewer sofisticado -> Formula 7 (WSP-brand).
• Regra transversal de batching: clonar hook+duracao de um Reel que JA performou antes de inventar formato novo.
Mockup adjustments
From the attached mockups
• Bios travadas verbatim para Dave e Felipe exatamente como nos mockups — nao reescrever.
• Highlights na ordem travada com icones que batem: Dave (Workflows, AI in Finance, Research, Prompts, Portfolio); Felipe (Guides, AI Finance, Prompts, Research, Tools).
• 9 grid covers verbatim por fundador, uniformes: fundador mid-gesture + titulo curto, touro visivel, set escuro; Dave dourado, Felipe limpo/levemente mais quente.
• Handles e numeros dos mockups: davewang.ai (48.7k), felipesinisterra.ai (28.7k).
• Subline Felipe verbatim 'Subscribe + grab the free guides (link below)' -> linktr.ee/felipesinisterra; subline Dave 'Free research workflows + the weekly prompt -> tap below' -> davewang.ai.
• Funil ManyChat: PDF hospedado ANTES da keyword ir ao ar; tabela source-of-truth de keyword unica (1->1->1->1->owner->status); zero overlap entre Dave e Felipe; WSP 'Second-Order' sem keyword.
Execution checklist
Do this, in order
1. Construir 1 HeyGen Photo Avatar por fundador das stills 2026-06-04 e travar (nunca regenerar).
2. Montar e fotografar o set travado (props + luz + vest) e congelar como referencia.
3. Aplicar bios verbatim + highlights na ordem + 9 grid covers por fundador.
4. Hospedar TODOS os PDFs dos lead magnets antes de qualquer keyword.
5. Montar tabela source-of-truth de keywords e ligar fluxos ManyChat (deliver -> 1 pergunta qualificadora com botoes -> rota newsletter + tag).
6. Testar cada keyword ponta a ponta (comentario -> DM -> PDF -> tag -> email) antes de postar.
7. Escolher 4 referencias provadas da Rip List por fundador a cada semana e clonar hook+duracao.
8. Escrever 4 scripts/fundador com hook na tela no 1o segundo + 1 keyword verde #16C784.
9. Filmar os 4 Reels em 1 sessao por fundador (set/vest/framing travados).
10. Burn captions palavra-por-palavra, 1 verde por beat, margem segura.
11. Postar SEG/QUA/SEX/DOM no horario fixo US-ET; Stories quase todo dia ao redor.
12. Rodape 'Educational only. Not advice.' em todo post e DM; nunca buy/sell/hold.
13. Aplicar gate de ads de 72h; passou -> ad nativo 9:16; falhou -> Content Radar.
14. Rodar lab de validacao (<=4-5 celulas ABO, hook-test primeiro) com budget aquecido devagar.
15. Logar tag por intencao e disparar nurture de 3 emails por segmento.
16. Manter WSP institucional em hold (2-3/semana) ate o estilo proprio ser fechado.
By Sept 2026, make "Wall Street Prompt" the default answer when a US self-directed investor searches (on IG/TikTok/YouTube, not just Google) "how do I analyze a stock with AI" — measured by branded-search lift + saves/shares per post + qualified comment-keyword DMs converting to the two founder newsletters (Dave 41k, Felipe 25k). The mechanism is not "more posts" but a small set of NAMED, recurring, keyword-optimized series that each own a search phrase and feed one funnel: free verified workflow (social) -> DM deliverable (ManyChat) -> newsletter -> WSP membership.
Upgrades
From → To
From angles to NAMED appointment series
from: 7 standalone validated angles published ad hoc across Week 1-4
to: 4 named, scheduled, keyword-owning series: (1) Dave 'AI x Markets in 60' (weekly Mon, news-beat, owns 'this week in AI and markets') built on the Rowan-Cheung cadence with a fixed intro/outro template; (2) Dave 'Copy This Prompt' (weekly Wed, prompt-as-asset -> keyword DM) the maestroprompts engine; (3) Felipe 'Build the Muscle' (weekly Tue, contrarian verification reframe) the Rachel-Woods frame; (4) WSP 'Second-Order' (biweekly Thu carousel, institutional) the jarvisinvest frame. Each gets a fixed cover lockup + episode number so the audience learns to expect it.
June 2026: recurring/episodic beats one-off; appointment content drives repeat-watch + follow, the signal short-form algos reward most. The angles are already series-shaped — naming + scheduling them is the highest-leverage change with zero new asset cost.
From hook-led to search-led captions/titles
from: Caption opens with the creative hook; on-screen text is the hook; filenames arbitrary
to: Every post leads its caption AND its YouTube/TikTok title with the literal search phrase, then the creative hook: e.g. 'How to analyze a stock with AI (the exact prompt) — I deployed $1.5B and use this to read an earnings call in 4 minutes.' Maintain a 12-phrase target-keyword map (analyze a stock with AI, read a 10-K with ChatGPT, AI earnings call summary, AI investing for beginners, ChatGPT stock research, etc.) and assign one phrase per series episode. Name exported files with the phrase.
Social-as-search now rivals Google for under-30 US users; relevance>reach. Keyword-rich first-line + title is free discovery that compounds, and finance terms have high commercial intent. Costs nothing, multiplies reach of the same asset.
From dead-drop DM to zero-party profiling loop
from: ManyChat sends the PDF + a 'soft route to newsletter'
to: Each keyword tags the subscriber by intent (PROMPT=workflow-builder, FILING/TENK=fundamental-research, BLUEPRINT=beginner). DM asks ONE qualifying question (beginner / self-directed / pro) before delivering, then routes into a behavior-triggered 3-email sequence matched to that segment. Pass the segment tag to the newsletter ESP.
Zero-party data + behavior-triggered email are explicit 2026 levers. The keyword already IS a declared interest — capturing it turns one DM into a segmented nurture, lifting newsletter->membership conversion without more reach.
From newsletter-as-endpoint to a micro-community layer
from: Funnel terminates at newsletter subscribe
to: Add a free, gated micro-community as the retention layer between newsletter and paid WSP: a 'WSP Prompt Vault' space (Discord or a private channel) where each week's prompt + a 'post your output, get it pressure-tested' thread lives. Comment-keyword VAULT delivers the invite. Seed it with the 7 existing lead magnets so it has day-1 value.
Micro-communities + relevance>reach are core 2026 plays; a community converts passive savers into retained, identifiable members and creates UGC (members posting their AI research outputs) that feeds more content. Read-only mandate respected — humans run the community, Claude only maintains assets/copy.
From 'separate YouTube' to a Shorts-from-long-form engine + LinkedIn bridge
from: Dave: YouTube long-form treated as a separate, lower-priority track
to: Make YouTube the hub for Dave: every long-form is cut into 3-5 Shorts (one beat each) that double as IG Reels/TikTok, each Short titled with a target search phrase and ending on a 'full breakdown on YouTube' CTA. Run weekly Community Posts to wake the 2.1k base, and post a LinkedIn native teaser that bridges his large LinkedIn audience to the new YouTube video.
founder-research shows YouTube-first DNA, an already on-brand green/matrix identity, and a 2.1k/41k sub-to-newsletter gap = the biggest owned-audience arbitrage. One long-form -> 5 search-optimized Shorts is the cheapest way to feed every short-form channel at once.
From 'learn-from creators' to a collab/seeding ladder
from: 5 creators mapped purely as reference to copy
to: Add a 3-tier collab ladder: (a) nano/micro AI-finance creators (aifinancetech, quantinsider.io, tradingcomdados once captured) seed them the prompt packs to feature; (b) peer-tier collab/guest swap (a joint 'AI vs analyst' challenge); (c) aspirational appearance target (Dave as guest on an AI-news show in the Rowan Cheung orbit). Start with tier (a) this phase.
Creator-driven/UGC + nano-micro collab is a named 2026 growth lever. Cold-start reach for the founder accounts is faster borrowed than earned; seeding real prompt assets (not asking for shoutouts) is the credible, finance-appropriate version.
From vague measurement to a rigor spec
from: 'Measure engagement', 'review retention', 'adjust if needed'
to: Define and instrument: (1) per-post target = saves+shares per 1k reach as the primary KPI (vanity views secondary); (2) A/B every post on ONE variable at a time (3s hook OR cover OR CTA), 2 variants, decide on 48h saves-rate; (3) per-keyword attribution dashboard (comments -> DMs -> subscribes -> segment); (4) reply-to-every-comment-<24h SLA (engagement is the ranking signal); (5) weekly winner-roll: top-saved format becomes next week's variation seed.
Engagement (saves/shares) > vanity reach and rigorous A/B + attribution are explicit 2026 requirements. Without a primary KPI and a one-variable A/B rule, 'double down on winners' is unactionable.
From unproven TikTok/X to a validated reuse standard
from: TikTok = 'reuse + more aggressive hook'; two X/TikTok angles sit unvalidated in the library
to: Hold all needs-validation TikTok/X angles out of the calendar until one real post is captured. Define a TikTok reuse standard from what we DO know: re-cut the Reel to a colder 1.5s hook, native text-overlay (not burned IG captions), trending-audio bed under the VO, and a TikTok-specific keyword title. Capture 5-10 AI-finance TikToks next agent-browser pass before scaling.
Over-produced/cross-posted clips underperform on TikTok in 2026; native pacing + native text + trending audio is required. Shipping the unvalidated X/TikTok angles risks burning a cold account on an unproven format — validate before scale.
From static compliance to compliance-as-positioning
from: 'Educational only. Not advice.' as a footer disclaimer
to: Make verification the visible product: a recurring on-screen 'VERIFY THIS' beat in every prompt video (show pulling the actual filing to check the AI), and a 'Sources' note on second-order/news posts. Keep the disclaimer, but lead with the verification ritual as the differentiator vs hype accounts.
In AI-finance, the trust gap IS the market (per the Felipe 'got burned by a confident wrong answer' pain). Showing verification on-screen converts the legal requirement into the reason to follow WSP over slop accounts — and is exactly Felipe's stated edge.
Series
Named appointment content
Weekly · Mon · Reel/Short
owns "this week in AI and markets"
Fixed intro/outro + episode number (Rowan-Cheung news-beat cadence)
Weekly · Wed · Carousel/Static
owns "copy this prompt"
Prompt-as-asset → keyword DM (maestroprompts engine)
Weekly · Tue · Reel/Carousel
owns "the mistake most people make with AI"
Contrarian verification reframe (Rachel-Woods frame)
Biweekly · Thu · Carousel
owns "second-order AI trade"
Institutional second-order thinking (jarvisinvest frame)
Aggressive plays
- LAUNCH MOMENT for Felipe's cold IG/YT (he has no IG/YT yet): a 5-day 'AI Investing Edge Challenge' — one named prompt per day (Mon-Fri), each a copy-this-prompt Reel with a unique keyword DM (DAY1...DAY5), capped Friday by the full 'If I started over' blueprint short. Pre-announce to his 25k newsletter + LinkedIn to inject day-1 followers into the new accounts so they don't launch to zero. Cross-promote on Dave's accounts.
- 'AI vs the Analyst' proof series: Dave runs his real research-assistant prompt live on a just-released 10-K/earnings call and races a stopwatch (4 min vs 40), showing the VERIFY step on-screen. This is the compliant version of the maestroprompts 'proof-of-result' hook — real, demonstrable, no return promise. Becomes the flagship hero asset and the thing collab partners want to feature.
- Comment-keyword surge week: concentrate the 7 ready lead magnets into a single 7-day 'Prompt Drop' week (one keyword/day), each post engineered for saves+shares, all routing into the new VAULT community. Turns the existing finished assets into one loud, attributable spike instead of a slow drip.
- Reddit/LinkedIn-group seeding (human-run, read-only mandate intact): repackage the FILING and TENK 10-K prompts as a genuinely useful text post in r/investing-adjacent and finance LinkedIn groups, value-first with the VAULT invite as the only ask. Borrowed-audience reach for the highest-intent finance crowd.
- Nano-creator prompt-pack seeding: DM the WSP prompt packs to 10-15 captured/near-lane AI-finance nano creators (aifinancetech, quantinsider.io, tradingcomdados) to use and credit — free assets, not paid shoutouts, which reads as credible in finance and creates UGC at the top of funnel.
- 'Same prompt, every quarter' evergreen hook: tie the Copy-This-Prompt series to the earnings-season calendar so each earnings week re-triggers the same high-intent search demand ('AI earnings call summary') with a fresh company — turning one validated asset into a recurring, calendar-driven traffic engine.
US-market notes
- • US self-directed retail (the idealClient set: 28-45 professionals, ex-burned-by-AI skeptics, beginners intimidated by jargon) searches IG/TikTok/YouTube as a how-to engine — lead with US-native search phrases ('analyze a stock with AI', 'read a 10-K with ChatGPT', not India/PT-BR framings). Jarvisinvest's captured posts are India-market recaps; copy ONLY the second-order STRUCTURE, swap to US tickers/events.
- • SEC/FINRA exposure is real for finance content to US audiences: the educational-only + no buy/sell/hold + no return-promise guardrail is non-negotiable and is also a credibility asset vs the hype lane. Keep 'not financial advice' visible; never imply personalized advice in DMs.
- • Earnings-season cadence (Jan/Apr/Jul/Oct) is the US demand calendar — schedule the Copy-This-Prompt and AI-vs-Analyst series to ride live earnings weeks for maximal search intent on real US names.
- • US short-form prime windows + reply-<24h matter for the US timezone audience; schedule the named series at consistent US-ET slots so 'appointment content' actually lands when the US audience is on-platform.
- • Dave's Harvard/$1B+ and Felipe's ex-Goldman/SoftBank are premium trust signals specifically calibrated to the US finance audience's respect for pedigree — keep them in first-line copy; they are the believability that lets a big time-compression claim land.
- • TikTok regulatory uncertainty in the US persists into 2026 — treat TikTok as a reuse/upside channel, not a primary bet; keep IG + YouTube as the owned core so the funnel survives any TikTok disruption.
Working
Keep
- ✓ The locked Shorts/Reels visual system (HeyGen avatar + dark studio + navy WSP vest + green word-by-word captions + 3s hook). Identical-recognition look is the engagement lever top creators use (Rowan Cheung, Sabrina) — do not touch it.
- ✓ The near-black + single-green + Montserrat static layout and the copy-this-prompt block. It is distinctive, premium, and already differentiated from generic AI-hustle slop.
- ✓ The ManyChat comment-keyword -> DM funnel with one UNIQUE keyword per asset and the 7 ready lead magnets (PROMPT, FILING, STACK, TENK, WORKFLOW, TIE-OUT, TOOLS). This is the proven conversion path (maestroprompts/Sabrina both run it live).
- ✓ The validated, exact-source-modeled content library (4 statics, 4 carousels, 4 shorts) and the FACT/INFERENCE/OPINION + 'no target, no recommendation' compliance guardrail. Compliance-by-design is a moat in finance, not a tax.
- ✓ Account roles: Dave = depth/authority (YouTube-first + IG/TikTok reuse), Felipe = practical/didactic (IG/Reels-first), WSP = institutional-neutral. The angle/format split (not seniority) is correct per founder-research.
- ✓ The real-credibility anchors used in copy (Dave: $1B+ deals, Harvard AI Fellow; Felipe: ex-Goldman/SoftBank, $1.5B). Never invent beyond these.
- ✓ The 'AI extracts/builds the read, you keep the judgment' house line — it is the brand's spine and ties every angle back to verification.
Gaps
Fix
- • NO NAMED EPISODIC SERIES. Current plan is one-off clips per angle. June 2026 reality: recurring named series beat one-offs (YT Shorts ~70B daily views reward repeat-watch behavior). The angles (e.g. 'This Week in AI x Markets') are series-shaped but unnamed, unbranded, and not scheduled as appointment content.
- • NO SOCIAL-AS-SEARCH DISCIPLINE. Captions/titles are hook-led but not keyword-led. Under-30 US investors search IG/TikTok/YouTube directly; there is no rule to put 'analyze a stock with AI', 'read a 10-K with ChatGPT', 'AI earnings call' literally in the first line of caption + on-screen + filename.
- • NO COMMUNITY / MICRO-COMMUNITY MOTION. Funnel ends at newsletter. No Discord/private group/Reddit/LinkedIn-group layer where saves convert to retained members and zero-party data. This is the 2026 relevance>reach lever and is entirely absent.
- • NO RETENTION / A-B / ATTRIBUTION RIGOR. Timeline says 'measure engagement' and 'review retention' but defines no metric targets, no A/B protocol (hook/cover/CTA), no per-keyword attribution beyond the keyword existing. No reply-<24h SLA despite engagement being the ranking signal.
- • DAVE'S YOUTUBE IS UNDER-LEVERAGED. founder-research shows YouTube-first DNA (39 videos, green/matrix identity already on-brand) but only 2.1k subs vs 41k newsletter + large LinkedIn. The GTM treats YouTube as 'long-form, separate' with no Shorts-from-long-form engine or LinkedIn->YouTube bridge.
- • NO CREATOR / NANO-MICRO COLLAB PLAY. We map 5 creators to LEARN from but never to COLLAB with. 2026 growth is creator-driven/UGC; no plan to appear on adjacent finance/AI creators or run nano-influencer prompt-pack seeding.
- • NO ZERO-PARTY DATA / BEHAVIOR-TRIGGERED EMAIL LOOP. ManyChat delivers a PDF and 'routes to newsletter' but there is no segmentation (which keyword = which interest) or behavior-triggered sequence off that signal. The DM is a dead-drop, not a profiling event.
- • FELIPE HAS NO VIDEO PROOF ASSET YET. He is IG/Reels-first but has noInstagramYet + noYouTubeYet in research — net-new channels. No 'launch moment' to create initial noise for a cold account.
- • TIKTOK & X ANGLES ARE UNVALIDATED. Two hooks + one short are status needs-validation (no capture). They sit in the library as if usable; risk of shipping unproven format.
- • NO 'PROOF-OF-RESULT' ASSET. maestroprompts' highest-converting hook is proof-of-result ('2M views in 7 days'). WSP has real assets (live prompt demos, before/after research speed) but no compliant proof artifact to anchor a hook — currently flagged 'swap in real metrics' with nothing to swap in.
Launch checklist
Everything needed before the soft launch
Next steps
Do these first
Missing assets
Blocking the launch
Launch timeline
Week 0 → Week 4
- Define bios (Dave/Felipe/WSP)
- Adjust bio links + central page
- Choose profile photos
- Choose video backgrounds (needs ref images)
- Lock Shorts/Reels visual standard
- Lock static-post layout
- Prepare first batch of content
- Publish first static posts
- Test first Shorts/Reels
- Repurpose to TikTok (swap captions)
- Start measuring engagement
- Validate hooks + CTAs
- Hold cadence
- Repeat winning formats
- Tune hooks
- Test comment/DM CTAs
- Turn best-responding content into variations
- Double down on best-performing formats
- Review Shorts/Reels retention
- Improve thumbnails/covers
- Adjust bio + link if needed
- Create new batches
- Scale winning themes
- Expand to new formats
- Feed Content Radar with new learnings
By account
Per-account focus + platforms
Instagram + Reels + TikTok (reuse) + YouTube long-form; LinkedIn as existing authority base.
Operator-grade depth. Turn strong content into consistent short videos. Lock bio/link/background/visual.
Instagram + Reels + TikTok (reuse) + LinkedIn base; Shorts as primary video format.
Practical, didactic, direct. Shorts-first. Lock bio/link/background/visual.
Instagram + institutional LinkedIn; TikTok only later if it makes sense.
Brand, product, cases, social proof, institutional education. Bio/link/layout.
By platform
Role + actions per platform
Bios
v2 — premium, non-cliché, per platform
I led $1B+ in deals before AI could read a 10-K in 30 seconds. Now I rebuild every research step as a prompt and show my work. Harvard AI Fellow. CEO @ Wall Street Prompt. Educational, never advice.
Short: AI x markets, from someone who's run $1B+ in deals. I publish the exact prompts I use to research faster. Educational only.
Direct: The prompts a real investor uses to read filings, pressure-test a thesis, and skip the busywork. New one most weeks. Free workflows below.
I closed $1B+ in deals the slow way. Now I show the prompt that does the first 80% in minutes. Harvard AI Fellow, CEO @ Wall Street Prompt. Steal the workflows.
Short: Ex-dealmaker, now I turn investing research into AI prompts. $1B+ in deals. Harvard AI Fellow. Not advice, just the workflow.
Direct: Watch me read a 10-K with AI, then copy the prompt. Investor lens, not hype. Educational only.
Most 'AI investing' content is a screen recording of a chatbot. This isn't that. I led over $1B in deals as a full-time investor, I'm an AI Research Fellow at Harvard, and I run Wall Street Prompt. On this channel I take the actual steps of institutional research, reading filings, building a thesis, stress-testing it, comparing tools, and rebuild each one as an AI workflow you can reproduce. Series you'll find here: AI Investing Workflows, Prompt Builds, Insider Takes, and Tool Tests for investors. New high-impact prompt most weeks in the newsletter (41k+ readers). Everything is educational only and never financial advice, no buy/sell/hold and no return promises.
Short: I led $1B+ in deals and now build AI systems for investing research. Harvard AI Fellow, CEO of Wall Street Prompt. Each video: a real workflow you can copy, end to end. Subscribe to invest better with AI. Educational only, never financial advice.
Direct: Real investor. Real research workflows. Rebuilt as AI prompts you can copy. Harvard AI Fellow, CEO of Wall Street Prompt. New build most weeks. Educational only, not financial advice. Newsletter, free course, and prompt library linked below.
I rebuild institutional research as AI prompts and post the receipts. Led $1B+ in deals. Harvard AI Fellow. CEO @WallStreetPrompt. One high-impact prompt every week, free. Not financial advice.
Short: Investor who led $1B+ in deals | Harvard AI Fellow | CEO @WallStreetPrompt | AI prompts for real research. Educational, not advice.
Direct: AI x markets for people who actually do the work. Filings, theses, tool tests, the prompt every time. Weekly prompt below.
CEO @ Wall Street Prompt | Investor, $1B+ in deals closed | AI Research Fellow, Harvard | I turn institutional research into AI workflows analysts can reproduce | 41k read my weekly prompt
Short: CEO, Wall Street Prompt | Led $1B+ in deals | AI Research Fellow at Harvard | Building AI research workflows for investors (educational, not advice)
Direct: I show investors the exact AI workflows I use for research | $1B+ in deals | Harvard AI Fellow | CEO, Wall Street Prompt
I'm Dave Wang. I led over $1B in deals, I'm an AI Research Fellow at Harvard, and I run Wall Street Prompt. Pick where to start: the weekly prompt (41k readers), the free course, my YouTube workflows, or the full prompt library. Everything here is educational, never financial advice.
Short: Dave Wang. Investor ($1B+ in deals), Harvard AI Fellow, CEO of Wall Street Prompt. I rebuild real research as AI prompts. Start with the weekly prompt. Educational only, not advice.
Direct: Choose your path: Weekly Prompt (free) - Free Course - YouTube Workflows - Prompt Library - Wall Street Prompt. Educational only.
Spent years at Goldman and SoftBank deploying $1.5B. Now I teach the AI-for-finance skills firms actually hire for, one clear guide at a time (92 published). Founder @ Wall Street Prompt. No jargon, no advice, just the playbook.
Short: Ex-Goldman, ex-SoftBank. I make AI for investing make sense, no code, no jargon. Founder @ Wall Street Prompt. Educational only.
Direct: The AI + finance guide I wish I'd had on day one. Plain language. Real steps. 92 guides and counting. Start free below.
I deployed $1.5B at Goldman and SoftBank. Here I undo the jargon and show the AI-for-finance move step by step. Founder @ Wall Street Prompt. Save the ones you'll use.
Short: Ex-Goldman & SoftBank. I explain AI for finance so you actually get it. No jargon. Founder @ Wall Street Prompt. Educational only.
Direct: If you skipped the finance background, start here. One AI skill, explained plainly, per video. Educational only.
I spent years at Goldman Sachs and SoftBank and helped deploy $1.5B, and most finance content still talks past the people trying to learn it. This channel fixes that. I take one AI-for-finance skill at a time, the same ones I cover across my 92 published guides, and explain it step by step so you can use it today, even if you have no finance or coding background. Expect plain-language breakdowns, real workflows, and the verification habits that separate useful AI from confident nonsense. Founder of Wall Street Prompt. Free guides and newsletter (25k+) linked below. Educational only and never financial advice, no buy/sell/hold, no return promises.
Short: Ex-Goldman Sachs, ex-SoftBank ($1.5B deployed). I teach AI for finance in plain language, no code and no jargon. Founder of Wall Street Prompt. Each video is one clear, usable skill. Educational only, never financial advice.
Direct: AI for finance, explained like a patient mentor. Ex-Goldman, ex-SoftBank. One usable skill per video, no jargon. Founder of Wall Street Prompt. Educational only, not financial advice. 92 guides + newsletter linked below.
I deployed $1.5B at Goldman and SoftBank, now I teach AI for finance without the jargon. Founder @WallStreetPrompt. 92 guides, 25k readers. Plain language, real workflows, never advice.
Short: Ex-Goldman & SoftBank ($1.5B deployed) | Founder @WallStreetPrompt | AI for finance, no jargon | 92 guides, free. Educational, not advice.
Direct: AI + finance, made usable. One skill at a time, no background required. Free guides below. Educational only.
Founder @ Wall Street Prompt | Ex-Goldman Sachs & SoftBank, $1.5B deployed | I make AI-for-finance skills usable for people without a finance background | 92 guides, 25k+ readers
Short: Founder, Wall Street Prompt | Ex-Goldman Sachs, ex-SoftBank ($1.5B deployed) | Teaching AI for finance in plain language (educational, not advice)
Direct: I teach the AI + finance skills firms hire for, no jargon | Ex-Goldman, ex-SoftBank | Founder, Wall Street Prompt
I'm Felipe Sinisterra. Ex-Goldman Sachs and SoftBank, $1.5B deployed, now Founder of Wall Street Prompt. I teach AI for finance in plain language. Start where you need: a free guide, the 92-guide library, my weekly newsletter, or the free course. Educational only, never financial advice.
Short: Felipe Sinisterra. Ex-Goldman, ex-SoftBank ($1.5B deployed), Founder of Wall Street Prompt. AI for finance, no jargon. Start with a free guide. Educational only.
Direct: Choose your start: Free Guide - 92 Guides Library - Newsletter - Free Course - Wall Street Prompt. No jargon. Educational only.
Where institutional-grade research meets AI you can actually reproduce. Founded by Dave Wang ($1B+ in deals, Harvard AI Fellow) and Felipe Sinisterra (ex-Goldman, ex-SoftBank). Workflows and prompts, verified and source-backed. Educational only, never advice.
Short: AI workflows for serious investing research. Built by people who've deployed billions. Verifiable, neutral, educational. Not advice.
Direct: The prompts, workflows, and checklists behind real research. Source-backed, no hype, no recommendations. Free starter pack below.
Wall Street Prompt | The standard for AI in institutional-grade research | Founded by Dave Wang (CEO, $1B+ in deals, Harvard AI Fellow) and Felipe Sinisterra (ex-Goldman & SoftBank) | Workflows, prompts, and verification, not advice
Short: Wall Street Prompt | AI research workflows for investors | Built by a $1B+ dealmaker and an ex-Goldman/SoftBank operator | Educational, verifiable, never advice
Direct: AI workflows and prompts for real investing research | Source-backed and verifiable | Educational, never financial advice | Wall Street Prompt
The standard for AI in institutional-grade research. Source-backed workflows and prompts, never recommendations. Founded by @davewangmia and @fasinisterra. Educational only.
Short: AI workflows for serious investing research. Built by a $1B+ dealmaker (@davewangmia) and an ex-Goldman/SoftBank operator (@fasinisterra). Educational, not advice.
Direct: Prompts, workflows, and verification for real research. No hype, no calls. Free starter pack below.
Wall Street Prompt is the standard for using AI in serious investing research. Founded by Dave Wang (CEO, over $1B in deals, AI Research Fellow at Harvard) and Felipe Sinisterra (ex-Goldman Sachs, ex-SoftBank). This channel publishes reproducible AI workflows, prompt builds, and verification methods, the kind of process real analysts use, with sources you can check. Nothing here is a recommendation, a price target, or a return promise. Everything is educational and built to be verified. Membership, newsletter, and the educational hub are linked below.
Short: Wall Street Prompt. AI workflows for institutional-grade investing research, built to be reproduced and verified. Founded by Dave Wang ($1B+ in deals, Harvard AI Fellow) and Felipe Sinisterra (ex-Goldman, ex-SoftBank). Educational only, never financial advice.
Direct: Reproducible AI research workflows, prompt builds, and verification methods. Source-backed, neutral, educational only, never financial advice. Founded by Dave Wang and Felipe Sinisterra. Membership and resources linked below.
Wall Street Prompt. AI workflows for institutional-grade research, founded by Dave Wang ($1B+ in deals, Harvard AI Fellow) and Felipe Sinisterra (ex-Goldman, ex-SoftBank). Start here: the free starter pack, the membership, the newsletter, or the educational hub. Source-backed and verifiable. Educational only, never advice.
Short: Wall Street Prompt. The standard for AI in investing research. Workflows, prompts, and verification, not advice. Start with the free pack. Educational only.
Direct: Choose your start: Free Starter Pack - Membership - Newsletter - Educational Hub - Case Studies. Educational only, not advice.
Bio link strategy
Where the bio link points + central page
Central page (or davewang.ai)
Central page destinations- • Newsletter (41k)
- • Free research workflows / templates
- • YouTube @davewang_ai
- • WSP membership
felipesinisterra.com
Central page destinations- • Newsletter (25k)
- • 92 guides library
- • Free templates
- • WSP membership
wallstreetprompt.com
Central page destinations- • Product / membership
- • Newsletter
- • Case studies / social proof
- • Educational hub
Static post layout
Instagram static / carousel standard
Template: Single 4:5 (1080×1350) tile, WSP near-black (#0B0E0C), Montserrat, ONE green accent (#16C784).
Visual: Heavy condensed all-caps headline, single green accent on the key phrase, thin green rule, monospace for any prompt block.
Background: Rarely — keep near-black solid; image only when it adds proof (chart/screenshot).
Graphics: Minimal — one chart or one prompt block per tile, never cluttered.
Carousel: For frameworks / prompt sets / step lists (premium 5-slide structure).
CTA: Comment-keyword (PROMPT / EDGE / FILING) → DM, or save-this. Always 'Educational only. Not advice.' footer.
Copy rules: Concrete > abstract · Real specifics only (no fake numbers) · No em dashes · Hook in line 1
Shorts / Reels art direction
Consistent look = recognition = engagement
Background: Dark cinematic home-office studio: blurred shelving left, warm practical lamp (amber) camera-right, cool-dark tones, shallow depth of field (bokeh). SAME backdrop for both founders.
Outfit: Dark navy/charcoal quarter-zip vest over a shirt + small light WSP logo on the left chest. Consistent every video.
Framing: Mid-chest crop, centered, eye-level, subject looking into camera. Cropped to 9:16 vertical, subject centered.
Talking head: HeyGen Photo Avatar per founder (Dave / Felipe) — same avatar + look every video for recognition.
Captions: Bold burned-in word-by-word captions, single WSP-green (#16C784) keyword highlight.
Hook (3s): First 3s: bold verbal + on-screen hook (curiosity / contrarian / outcome). No warm-up.
Final CTA: Comment-keyword → DM, or follow-for-more.
Duration: 45–75s (≤90s ceiling).
Pace: Fast cuts, one idea per beat, no dead air.
TikTok reuse: Reuse the Reel; swap caption + first-frame hook for a more aggressive TikTok angle; measure retention.
Reference set by founder's two avatar stills. Top creators keep background, outfit, framing and language identical across videos — this recognition is the engagement lever. Same standard applies to Dave and Felipe (same set, different avatar).
Viewer comments the keyword on the post → ManyChat auto-DMs the deliverable (PDF/link) + routes to the newsletter. Keyword must be UNIQUE per asset or ManyChat cross-wires the funnel.
Boundary: WSP creates the ManyChat account and connects Instagram (Meta OAuth) — Claude does NOT create accounts, log in, or connect. Claude maintains the keyword→asset map, DM copy, and status here.
Keyword registry
One unique keyword per asset — collisions cross-wire the funnel
| Keyword | Account | Deliverable | Asset | Status |
|---|---|---|---|---|
| PROMPT | Dave | Research-assistant prompt + verification checklist (PDF) | ready | ready |
| FILING | Dave | 25 AI prompts for reading a 10-K (PDF)PDF now delivers all 25 prompts. | ready | ready |
| STACK | Dave | 10 AI tools for 10-K analysis — full breakdown (PDF)Slides are generic categories — confirm the 10-tool deliverable exists (QA flag). | ready | ready |
| TENK | Dave | Exact 10-K reading prompt structure (PDF)Use 'TENK' not '10K' — ManyChat keyword matching is cleaner with letters. | ready | ready |
| WORKFLOW | Dave | One-page research workflow (PDF)Collision resolved: TOOLS keyword now covers the free-tools post (DYuXBOKjB6b). | ready | ready |
| TIE-OUT | Dave | Model-review checklist (PDF)Confirm checklist asset + DM owner before the keyword ships (QA flag). | ready | ready |
| TOOLS | Dave | Free AI analyst tool stack (PDF)Split from WORKFLOW to keep keywords unique. | ready | ready |
Setup checklist
WSP connects ManyChat; Claude maintains the map
Rules
Non-negotiable
Use Meta ads as a CONTENT VALIDATION LAB, not a distribution firehose. The goal in this phase is not conversions, it is to buy statistically clean signal on which organic hooks, formats and founders actually hold attention and pull profile action so the winners get scaled organically and the losers get killed before they eat the calendar. Run on a NEW Meta ad account warmed slowly (low daily budgets, 3-5 day learning windows, never edit ad sets mid-learning). Every ad is an existing or near-final organic asset (the validated Reels/carousels modeled on maestroprompts copy-this-prompt, sabrina_ramonov blueprint, the.rachelwoods contrarian, jarvisinvest second-order, rowancheung news cadence) run essentially un-retouched: native 9:16, 3s hook, WSP-green keyword captions, no over-produced ad polish (June 2026: organic-mirroring creative outperforms ad-looking creative). Two distinct objective tracks because they measure different things: (1) ENGAGEMENT / video-views objective for cheap top-of-funnel signal on hook strength and 3s/retention curves; (2) a small TRAFFIC / profile-visit + Instagram-promote layer to test whether attention converts to the profile action that drives the comment-keyword to DM funnel (PROMPT, FILING, WEALTH/BLUEPRINT, STACK, TENK, WORKFLOW, TOOLS). Read-only social discipline holds: ads only amplify our own posts, we never recommend auto follow/like/comment, and every creative keeps the educational-only framing with no buy/sell/hold or return language. Test ONE variable per cell (hook OR format OR founder OR audience), label everything, and let organic carry the funnel once paid identifies the winner. Per-account split of spend: Felipe and Dave carry most of the paid-test budget (short-form is the validation surface); WSP runs one small institutional cell only.
A/B tests
Ads validate content, not just distribute
| Hypothesis | Variant | Audience | Metric | Decision |
|---|---|---|---|---|
| Proof-of-result hooks (maestroprompts style) beat utility hooks for 3s hook-rate and hold-rate in the AI+finance lane. | Same Felipe Reel, two first-3s hooks: A = outcome/curiosity ('If I started over in AI investing, here is the exact blueprint'), B = pure utility ('3 AI prompts that read a 10-K for you'). Identical body + caption + CTA. | COLD: US, 25-45, interests AI tools, ChatGPT, investing, personal finance, fintech; broad-ish, no stacking. Same audience both ads, ABO so budget is forced even. | Primary: 3s hook-rate (3s plays / impressions) and 50%+ video hold-rate. Secondary: cost per 3s view, profile visits, follows. | SCALE the winning hook if its hold-rate is >=25% higher at <= equal cost/3s-view; KILL the loser hook in this format; ITERATE the marginal case (within 15%) by re-cutting the weaker hook once before retiring it. |
| FORMAT TEST: a talking-head Reel (sabrina/rachel style) validates faster and cheaper than the static carousel for the same idea. | Same core idea ('the mistake most people make with AI in finance') as A = 60-75s talking-head Reel vs B = 5-slide carousel. Same hook line, same comment-keyword CTA (BLUEPRINT). | COLD US 25-45 AI+investing interest stack, identical to test 1's audience for comparability, ABO even budget. | Cost per profile visit + cost per save/share (saves/shares are the engagement-quality signal per June 2026 relevance-over-reach); secondary cost per result and CTR. | SCALE the cheaper-per-profile-visit format as the default for that idea across the calendar; ITERATE the losing format with a stronger cover/first-frame once; KILL only if cost per profile visit is >2x the winner after full learning window. |
| FOUNDER TEST: Felipe (accessible/didactic) beats Dave (operator-depth) on cold IG short-form for hook-rate and follow cost; Dave wins on engagement DEPTH (saves/comments per reach). | Same concept, same script skeleton, two founders/avatars: A = Felipe cut (jargon-free framing), B = Dave cut (operator-grade, $1B+/Harvard credibility, YouTube-feeder angle). Hold format + audience constant. | COLD US 25-45 same interest stack; run as two ad sets so each founder gets a clean even-budget read (accept controlled audience-overlap risk at this small scale). | Follow cost + 3s hook-rate (Felipe lens) AND saves+comments per 1k reach (Dave lens). Track both, do not collapse to one number. | SCALE each founder to the objective they win (Felipe -> top-of-funnel reach/follows; Dave -> deep-engagement + YouTube feeder). Do NOT kill the 'loser' globally; assign by job. ITERATE if both lose to a neutral WSP cut. |
| COLD-vs-REMARKETING: warm video-viewer/engager retargeting drives the comment-keyword DM funnel far more efficiently than cold, validating the funnel mechanic itself. | Identical best-performing creative + identical comment-keyword CTA (PROMPT / FILING). Run cold vs a custom audience of 50%+ video viewers + IG engagers (last 14-30 days). | A = COLD US interest stack. B = REMARKETING: video viewers (ThruPlay/50%) + IG profile engagers, last 14-30d. Only activates once the audience accumulates enough size (~wk 2-3). | Cost per profile visit and (funnel proxy) lift in keyword-comment volume / DM opt-ins on the boosted post vs an unboosted control post of the same asset. | SCALE remarketing as the standing always-on layer if cost per profile visit is materially lower AND it lifts keyword-comment volume vs control; if cold matches remarketing efficiency, KILL the separate remarketing spend and reinvest in cold prospecting; ITERATE audience window (14 vs 30d) if signal is thin. |
| INSTITUTIONAL CELL: a WSP brand-led asset (no founder face) can hold attention and validate paid demand in the AI+finance lane. | One WSP cell: A = founder-fronted version of a second-order/jarvis-style idea vs B = WSP brand-only version (no face, institutional caption, 'Educational only. Not advice.' footer). | COLD US 28-50, slightly older, interests investing/fintech/professional-finance + lookalike-seed once any engager list exists; single small ABO cell at the 12% budget allocation. | Hold-rate + saves per reach + profile visits; watch comment sentiment for compliance (no advice-seeking we cannot answer). | SCALE the WSP-brand format only if it clears the founder cut on saves/profile-visits at <=1.25x cost; otherwise KILL brand-only paid and keep WSP organic-only, routing paid budget back to Felipe/Dave. |
Metrics tracked
Short term
Before a formal community
- • Pre-community phase = run a 'belonging-before-platform' play. No Discord/Circle yet. The community lives inside comments + DMs + the keyword funnel until you have 2,000+ engaged followers per founder account. Building a room before you can fill it kills the room.
- • Reply to EVERY comment within 24h (target <2h in the first 90 min after posting — IG weights early engagement). Founders (or a ghost-writer in their exact voice) reply, never a generic brand voice. Pin one reply per post that adds a second insight, not 'thanks!'.
- • Turn the ManyChat keyword DM into a relationship, not a drop. After delivering the PDF, the DM asks ONE qualifying question ('Are you using AI for your own portfolio or for client/research work?'). Two-tap reply = zero-party data + a reason to DM back. This is your proto-community CRM.
- • Run a named recurring comment ritual that creates in-group identity: Dave posts a weekly 'Stress-test this' — drops one AI research output and asks followers to find the flaw in the comments. Best catch gets pinned + named. Manufactures belonging with zero infrastructure.
- • Seed micro-belonging by naming the audience. Rachel Woods segments with 'Save this if you're an AI Operator'. Adopt a WSP equivalent: Dave's people = 'operators', Felipe's = 'the desk'. Use the label in captions + DM copy so members self-identify before any platform exists.
- • Felipe (no IG/YT yet): warm-start the audience from his 25k newsletter + LinkedIn. First 2 weeks of IG = 'reply to me' prompts in Stories (polls, question stickers, 'comment your stack') to manufacture the early-engagement signal a cold account lacks.
- • Build a private 'founding members' soft-list NOW via the DM funnel: anyone who completes the qualifying question gets tagged 'early'. When the real community launches you DM these people first — a warm 200-person seed beats a public launch to silence.
- • WSP brand account: run belonging through saves + shares, not chat. CTA = 'Send this to the analyst on your team who still reads 10-Ks by hand.' Share-to-a-colleague is the institutional version of community in the pre-platform phase.
Medium term
When + where
- • Newsletter-first is the spine. Dave (41k) and Felipe (25k) already own owned-audience channels with behavior-triggered potential — do NOT split attention into a chat platform until a single account clears ~2,000 highly-engaged IG followers AND a repeatable weekly ritual is proven for 6+ weeks.
- • Sequence: (1) Newsletter is home base now. (2) Add a real-time community layer ~Week 8-12, only on the account with the strongest reply/DM volume (likely Dave first given operator depth + existing YouTube community tab). (3) WSP brand community comes LAST and is gated to paying members.
- • Platform for the founder community: Discord for Dave (operator/builder audience lives there; threads + voice AMAs + bot-gated channels fit 'stress-test the prompt' culture). Skool or Circle for Felipe (didactic, course-like; his 92 guides + free course map cleanly to a cohort/classroom layout; less dev-native audience).
- • WSP institutional community: gated Circle or a private LinkedIn group tied to membership — NOT Discord (compliance + institutional audience won't tolerate a gaming-coded tool). Keep it neutral, educational, structured channels only.
- • Trigger to launch a chat platform: 3 conditions met simultaneously — (a) 50+ inbound DMs/week you can't scale 1:1, (b) followers explicitly asking 'is there a group?', (c) a proven weekly ritual people show up for. Launching before all three = a ghost-town server that signals weak demand.
- • Behavior-triggered email is the connective tissue: keyword-DM completion → tagged email sequence (operators vs beginners vs institutional). This runs 'community' economics on owned infrastructure for months before paying for Circle/Skool.
- • Telegram and Slack are NOT recommended. Telegram skews crypto-signal (compliance risk + wrong association for an educational finance brand). Slack has no discovery, weak async-ritual fit, and reads as B2B-internal, not creator-community.
- • Monetization-readiness: when WSP membership launches the gated community, the founder Discord/Skool becomes the free top-of-funnel and the WSP-gated Circle becomes the paid tier. Free founder rooms feed the paid brand room — do not gate the founder rooms early or you choke top-of-funnel.
Weekly rituals
- • Dave — 'Stress-Test Tuesday' (weekly): post one real AI research output (a prompt result, a 10-K summary, a screened thesis) and ask operators to find the flaw in the comments. Best catch pinned + credited. Builds in-group identity, generates UGC, reinforces verification-over-hype. Educational only.
- • Felipe — 'Desk Drills' (weekly): a 60-90s didactic walkthrough of ONE AI-for-finance skill (e.g. 'how to make Claude label FACT vs INFERENCE'), ending with a comment-keyword to get the exact guide. Mirrors his 92-guide didactic DNA; trains the 'build the muscle' message (Rachel Woods reframe).
- • Both — 'This Week in AI x Markets, 60s' (weekly reel): fixed intro/outro template, one sourced AI-finance news item filtered through an investor's lens (Rowan Cheung cadence, finance-relevant topic, never off-lane robot news). Follow CTA. Creates appointment viewing.
- • Monthly live AMA: Dave on Discord voice (operator Q&A on workflows), Felipe on Skool/Circle live (didactic teardown of a member's AI finance question). Recorded, clipped to Shorts/Reels — the AMA feeds a week of short-form. Educational framing, no personalized advice.
- • 'Build-along' monthly thread (community platform, once live): founder ships a research workflow live across a week; members replicate it on their own tickers/sectors and post results. Converts passive followers into contributors and produces testimonials/UGC for the brand account.
- • Newsletter 'Reply of the Week': feature one subscriber reply/question in the next issue (zero-party data + belonging + makes replying a habit, which fuels behavior-triggered segmentation). No advice given — only the question + an educational answer.
- • WSP brand — 'Send-to-your-desk' weekly share prompt: one institutional-grade insight per week explicitly designed to be forwarded to a colleague/analyst. Share is the institutional community signal; track shares/saves as the KPI, not comments.
Lead magnets
- • Dave — 'The Operator's Verification Checklist' (community-gated upgrade of TIE-OUT): the public PDF is the teaser; the full annotated version + the prompt library that generated it lives behind joining the founding-members DM list. Qualifies serious operators, not freebie-hunters.
- • Felipe — 'The 7-Day AI Finance Desk Starter' (email mini-course): one didactic email/day, each ending in a reply prompt. Behavior-triggered: completing day 7 = invite to the cohort community. Maps to his free-course DNA and segments the truly engaged. Educational only.
- • Felipe — 'If I started over in AI + finance: the 10-step blueprint' (PDF, Sabrina Ramonov model reframed): keyword BLUEPRINT. Numbered, accessible, ex-Goldman/SoftBank framing, 'not a get-rich claim' disclaimer. High-share, beginner-qualifying top-of-funnel.
- • Both — 'The Second-Order Map' worksheet (Jarvis Invest model): a fill-in template walking an investor through 'the obvious AI trade → but then what?' for any sector. Keyword EDGE. Attracts thinkers, not tip-chasers; pressure-tested against filings, educational only.
- • WSP — 'The AI-in-Markets Skills Map' (Rachel Woods DZH9U3piKaI model): the AI+finance skills firms are hiring for, mapped to free resources. Save + share CTA, institutional-neutral. Qualifies the 'upskilling analyst/team' member the brand community wants.
- • WSP — 'Institutional AI Research Standards' one-pager: the neutral, compliance-safe framework (sourcing, FACT/INFERENCE labeling, human review) that defines the WSP standard. Gated to the paid-member community as the anchor document — the canonical thing people join to get.
- • Community-exclusive 'Prompt Drop' (members only, recurring): one new vetted research/analysis prompt per week, delivered only inside the community platform. The recurring reason to STAY a member — distinct from one-off acquisition magnets. Each prompt ships with limits + educational disclaimer.
Profile verification
Raise approval odds
- • Lock the real-name handle and full name to match his government ID exactly (David Wang / Dave Wang as on ID) before applying — name mismatch is the #1 rejection cause.
- • Set the profile photo to a clear real-face headshot (the HeyGen avatar is fine for content but use a true face photo on the profile for identity matching).
- • Complete bio, link (davewang.ai / central page) and category (Entrepreneur / Public Figure) so the profile reads as complete and notable.
- • Enable 2FA and confirm the email/phone are current — security posture is checked.
- • Build corroborating public proof of notability: Harvard AI Fellow, CEO Wall Street Prompt, press/podcast mentions — link these from the site so reviewers can verify the real person.
- • Apply for Meta Verified in the IG app and submit the government ID that matches the name; keep the account compliant (no advice/return claims) so it is not flagged.
- • For YouTube: pursue the handle checkmark only after 100k subs OR via Google identity verification; meanwhile cross-link LinkedIn/site to establish the canonical Dave Wang identity.
Meta Verified is a paid subscription (no follower minimum) but still requires: account in Dave's real legal name matching a government ID, a profile photo showing his face, prior posting history, and 2FA enabled. Authenticity, uniqueness (one real person, no impersonation), completeness (bio/photo set) and notability/security checks raise approval odds and protect against takedown. YouTube handle verification requires 100k+ subscribers OR a Google identity check; Dave is ~2,140 subs, so the YT checkmark is a later milestone.
- • Claim the canonical handle in his real legal name (matching government ID) on IG + YouTube simultaneously to prevent impersonation/squatting.
- • Warm-start the account from his newsletter + LinkedIn so it has real followers and engagement history (not a cold shell) before applying.
- • Post consistently for at least 4-6 weeks to build authentic posting history reviewers can see.
- • Set a real-face profile photo and a complete bio with verifiable credentials (ex-Goldman/SoftBank, Founder WSP) and link felipesinisterra.com.
- • Corroborate notability externally: keep site, LinkedIn and any press/interviews linked and consistent so the real-person identity is cross-referenceable.
- • Enable 2FA, set category to Public Figure / Entrepreneur, then apply for Meta Verified with the matching government ID.
- • Maintain compliance (educational only, no advice/returns) to avoid post-verification flags or removal.
Same Meta Verified mechanics (paid; real legal name matching ID, face photo, posting history, 2FA). Felipe's risk is a brand-new IG/YT account reading as low-trust, so the priority is establishing authentic activity and corroborating notability (ex-Goldman, ex-SoftBank, $1.5B deployed, 25k newsletter, 92 guides) BEFORE applying. Do not apply on a 2-week-old empty profile.
- • Ensure the IG handle, legal entity name and wallstreetprompt.com domain all match the business registration documents — inconsistency is the top rejection cause for business verification.
- • Set up and verify a domain email (e.g. team@wallstreetprompt.com) — business verification leans on domain ownership.
- • Complete the profile fully: logo, neutral institutional bio, site link, business category, contact info — a complete profile reads as established.
- • Establish external notability: press mentions, founders' credentials linked from the brand, consistent name/presence across the web so reviewers can corroborate the organization.
- • Apply for Meta Verified for Business and upload the legal business documents (incorporation/registration) matching the name and domain.
- • For LinkedIn: claim the Company Page and verify it via an admin whose personal profile is identity-verified and whose email is on the wallstreetprompt.com domain; keep founder profiles linked as employees to reinforce the entity.
- • Keep all brand content institutional-neutral and educational (no advice, no return claims, no em dashes) so the verified entity is never flagged for compliance.
Brand verification differs from personal: Meta Verified for Business requires legal business documentation (registration/incorporation docs), a matching business name and domain, and an official email on that domain, plus a complete, established presence. Brand 'notability' = external coverage and cross-web consistency, not follower count. LinkedIn requires a verified Company Page (admin tied to a verified personal profile + confirmed company-domain email).
LinkedIn announce posts
Announce new platforms without looking needy
Open with a concrete result or insight from a recent build (e.g. the Similarweb-as-hedge-fund prompt). One short paragraph on why text undersells it — 'you have to watch the model reason through a 10-K to get it.' Then a single soft line: 'I'm now posting these builds as 60-sec videos on IG and YouTube — same operator depth, just on camera.' Keyword-rich for LinkedIn search: 'AI equity research,' 'prompt for 10-K analysis,' 'AI investing workflow.' No 'please follow,' no link-dump. One destination max (central page). Educational-only footer. No em dashes.
Start with a reader question or DM he actually gets ('can you just show me where to paste the prompt?'). One paragraph: written guides are great for reference, bad for watching someone do it. Soft pivot: 'So I'm putting the simplest version on Instagram and Reels — no code, no jargon, one workflow per video.' Search keywords: 'AI for investing,' 'AI finance for beginners,' 'no-code AI investing.' Tease a named series (see migrationPlay), not a generic 'follow me.' One link. 'Not advice, educational only.' No em dashes.
Short, declarative. 'Wall Street Prompt's educational workflows now live in more places.' Bullet the channel map (Dave: YouTube + IG for operator-grade builds; Felipe: IG + Reels for accessible playbooks; newsletters as the written home). Position the new platforms as serving different learning styles. Keyword-rich: 'AI in markets,' 'institutional AI workflows,' 'AI investing education.' One link to the central hub. Neutral, no hype, no return claims, educational-only. No em dashes.
Add a fixed 2-line P.S. block to the bottom of each founder's existing newsletter (not a dedicated 'follow me' blast — that reads needy). Dave's: 'P.S. I'm now recording these prompt builds on camera. If you'd rather watch the 10-K teardown than read it: [central link].' Felipe's: 'P.S. New this month — the simplest version of each guide, in 60 seconds, on Instagram. Start with the [series name]: [central link].' Reframe the migration as a format choice for the SAME content the subscriber already opted into, so it lands as a service upgrade, not an ask. Rotate the specific artifact referenced each issue so it never feels like the same plug.
Soft, repeatable, end-of-video and end-of-caption line tied to a real reason to follow — never 'follow for more.' Dave: 'New build every week. If you want the next teardown, the prompts live where I post them.' Felipe: 'One workflow, one video, every [day]. Save this one and the series picks up from here.' The lever is the named episodic series (curiosity + continuity), not the founder. Pair with the comment-keyword ManyChat funnel already built (PROMPT / FILING / WEALTH-style) so the CTA delivers an asset AND earns the follow as a side effect. Keep it under one sentence, no exclamation stacking, educational framing.
One central link per founder used everywhere (LinkedIn bio, newsletter P.S., IG bio, YouTube about) — do NOT scatter platform-specific links across posts (looks needy and fragments attribution). Dave: a clean central page (davewang.ai or a /links page) where the FIRST and largest destination is 'Watch the latest build (YouTube/IG),' newsletter second, WSP third. Felipe: felipesinisterra.com gets a new top-of-page module 'Watch the guides (Instagram/Reels)' above the 92-guide library. Tag every outbound link with UTMs (source=linkedin/newsletter, medium=announce, campaign=migration-2026) so you can attribute which existing-audience channel actually converts to the new platforms — zero-party + behavior-triggered follow-up from there. The central page is the single hub; the new platforms are the hero module on it for the migration window, then settle into the normal stack.