Confidential · Strategy Overview

The Growth Machine

A viral game that turns AI's disagreements into verified truth, and the engine that sells it.

No other company in the world runs this many self-reinforcing flywheel loops. 3 frontier models (Anthropic's Opus, OpenAI's o3, and Google's Gemini) reviewed the field independently and couldn't find a competitor running more than 3. We run 9.

▶ It's all live: see every feature in the real product ↗

The Flywheelbadges & referrals feed new searches back to the top ↻
1

Disagreement

3 AIs answer every search; ~20% disagree. Free, renewable raw material.

2

Resolution

A verified hobbyist or expert settles the clash and gets paid for it.

3

Verified truth

Every resolution becomes proprietary, checkable data we own.

4

Revenue

Consumers, companies & AI labs all pay for that same verified truth.

5

Reinvest

Revenue recruits more hobbyists & experts → more clashes caught → repeat.

↻  Badges & referrals feed new searches back into the top
The 1 number that drives everything

The master signal

1 computed score, how much the AIs agree on any given question, powers every product, every viral loop, and most of the revenue that flows from them. Read one way it shows where to deploy human expertise. Read the other way it shows what's been verified and can be sold.

← Divergence pole: where AIs break
Virality + money + escalation
  • Raw material for viral games and drama
  • Triggers the hobbyist/expert escalation offer
  • Funds AI-disagreement bounties
  • Turns searchers into experts (Inbound hobbyist/expert trigger)
Convergence pole → where AIs get fixed
Trust + corpus + making AI better
  • Human-confirmed answers (crowd determines accuracy, not us)
  • The verified-truth corpus: 1 asset, 5 products
  • "Google Trends for where the AIs disagree" data layer
  • Closes the tagline: fix AI together
1 axis · 2 productized engines · 1 shared engineering dependency · 1 category story to coin
The one idea

AI is improving faster than the labs can keep it accurate. The fix is mass crowdsourcing: made viral.

3 AIs answer every search. When they disagree, that clash becomes a game people share, and a job a hobbyist or expert gets paid to settle. Every settled clash becomes verified truth we can sell. The machine that runs this loop is what we're building, and what you'd own as the GTM engineer running it.

The whole system on one screen

The 9 loops, at a glance

The core cycle above has 5 stages. What makes it a machine and not a funnel: 9 loops that each close back on it. 8 are reactive: they fire off a search, a clash, or a demand signal. 1 is proactive: it runs before anyone asks. Each line states the feedback that closes the loop.

1
Demand bounties
Pledges signal demand → the next expert is nudged to answer → more answers → more pledges.
2
The answer is the share-card
Every resolved answer is a badge billboard → free acquisition → more searchers → more answers to share.
3
Searchers become experts
Repeat searchers already know the topic → recruited to answer → supply grows out of demand, no outbound.
4
"Trending now" demand feed
Every return visit sharpens the feed → experts show up where demand is unmet → more resolved → more visits.
5
Peer referral
Everyone who earns has peers who'd earn too → referral bonuses → supply self-grows, coverage widens.
6
Peer review → reputation & debate
Experts review each other → reputation updates, and public disagreements become viral debate objects.
7
All-👎 → "what did they miss?"
Rejecting every AI answer is the highest-signal moment → an inbound expert audition at the cheapest point.
8
The advisor
Every conversation teaches it the user → sharper nudges → more participation, and it speeds loops 1–7.
9
The notebook  proactive
The only loop with no trigger: banks knowledge before anyone asks → sell (de-identified) → earn → bank more. Net-new supply, broader corpus.

Loops 1–8 harvest knowledge the moment a search or clash triggers it. Loop 9 banks what an expert already knows. Together, no unmet demand goes unanswered, and no expert's knowledge sits unsold. The rest of this page details each one.

Loops 1–7: the connections

The first 7 loops are connections: each turns a product moment into a supply or demand lever, and feeds the next. All 7 fire reactively, off a search or a disagreement. These are what make this a compounding machine, not a one-directional funnel.

Loop 1
Demand bounties: pledge to unlock a hobbyist or expert
When searchers hit a disputed question, they can pool interest to fund a human answer. Interest pledges only for now, no money captured yet. Each pledge is a demand signal that goes straight to the "your interests trending now" feed and nudges the next hobbyist or expert to write the answer.
Loop 2
The answer is the share-card
Every resolved answer becomes a shareable card carrying the answerer's badge. Each share is a billboard for their credential and a free acquisition card for the platform. They have every incentive to post it; every post is a free ad that refills the top of the funnel.
Loop 3
Searchers become hobbyists & experts
Someone who searches the same question 5 times probably knows the answer. The funnel catches them: search → vote → dispute → "want to answer this?" People who already know a topic are the cheapest contributors to recruit, no outbound required.
Loop 4
Every return visit feeds "your interests trending now"
Continuity (saved searches, follow-up questions, return visits) is the signal that demand is real and unmet. "Your interests trending now" aggregates these patterns anonymously and shows hobbyists & experts exactly where to show up. The more people search, the smarter the feed gets.
Loop 5
Hobbyists & experts recruit their peers
Anyone who earns here has colleagues in the same domain who'd earn here too. Referral bonuses turn the supply side viral: each one recruits the next, expanding the coverage map without us lifting a finger. The supply side self-grows.
Loop 6
Peer review as reputation and synthesized debate
Hobbyists & experts review each other's answers. That peer review is how reputation scores update, and when peers publicly disagree, their debate becomes a structured, crowd-scored "synthesized debate" that's itself a viral object. Disagreement compounds at every layer.
Loop 7
Inbound hobbyist/expert trigger: all-👎 → "what did they miss?"
When a user rejects every AI answer, that's the highest-signal moment in the funnel: they have an opinion the AIs missed. 1 free-text box ("what did they miss?") simultaneously captures a gripe, a demand signal, and an inbound hobbyist/expert audition. The cheapest contributor-sourcing funnel that exists.

No company currently combines all of it: a free AI search surface, a viral disagreement game, a human expert marketplace, credentials verified by the companies that hire them, wired directly to paying companies, an auto-generated "your interests trending now" demand signal, and a disagreement corpus sellable to AI labs. Loops 1–7 are what make the whole far greater than the parts, and two more run on top of them: the advisor (Loop 8, next) and the notebook (Loop 9), the only proactive loop.

Loop 8: the advisor · rolling out now

1 face that runs every loop

Now 1 AI advisor greets every user and pushes them into whichever loop fits: support, feedback, notebook, matchmaking & real disputes to solve. It's the 8th loop in its own right, and because it also feeds all 7 others, it's the operator layer that makes the rest spin faster.

Its own compounding cycle: every conversation teaches it more about the user → its next nudge (greeting, match, dispute-to-solve) lands sharper → that drives more participation & more conversations → it learns even more. Data → personalization → engagement → more data. That's a real flywheel in its own right, and it happens to also accelerate the other 7.

Greeter
Greets every new visitor, so fewer bounce
Confused & first-time visitors get a proactive, interest-tailored "here's what this site does for you." Memory travels across sites: site #2 says "welcome back, this one's different," never "hi, you're new" again. Fewer people bounce off a cold search box → the top of the funnel refills itself.
1 surface
Every thumb, gripe & idea becomes data
Absorbs the old 👍 / 👎 / 💡 buttons into 1 chat. Each reaction is now structured signal, and a real suggestion files straight into the founder's review queue while the advisor is still talking to you.
Matchmaker
"Would anyone pay me for this?"
The conversational front-end to the 2-sided marketplace. A user says what they know → the advisor matches it to a paying buyer, or asks which companies they'd target (free lead-gen either way). Every exchange also reads their expertise → raises their standing.
Router
Hands you a real fight to win
Surfaces a live AI disagreement in your wheelhouse: "an AI is claiming peaches are apples; want to settle it?" You answer → you're featured that day with your badge attached → that share refills the funnel & recruits the next solver.
Loop-closer
Tells you your answer landed
Reports back: "your answer was seen by N people, X thumbs up." Solving stops being a one-shot and becomes a measurable status event → you come back to solve more.
Notebook coach
Builds your sellable knowledge with you
Coaches you to add your links, work & know-how → more sellable notebook entries → a higher track record standing and more verified-truth corpus we can sell. It turns "I'll do it later" into a filled profile.

The advisor turns passive UI (buttons you have to hunt for, a blank box that scares people off) into an active guide that nudges every user 1 rung deeper into the loop that fits them. 1 persona, traveling every site; depth & humor flex to the person, not the paint.

Loop 9: the notebook · rolling out now

The one loop that runs before anyone asks

Loops 1–8 all fire when a search or a disagreement triggers them: they harvest human knowledge reactively, downstream of the funnel. Loop 9 is the only one that banks what an expert already knows, before anyone asks: a private "second brain" they fill for their own desktop needs, that turns into sellable supply. It's the only net-new supply source on this page.

Why it's a distinct loop, not a repackage: it needs no search, no clash, no demand signal to produce supply. Contribute → sell the de-identified knowledge → earn → contribute more. And the corpus it builds is broader than resolved disagreements: an expert's whole domain, which is the coverage AI labs actually pay for. Knowledge → money → more knowledge.

Proactive supply
The only loop with no trigger
Every other loop waits for a search, a clash, or a demand pledge. This one banks an expert's existing knowledge cold: knowledge that was never a disputed question. That's a supply source none of loops 1–8 can reach.
Real tool first
Useful on day 1, offline & alone
A genuinely good personal knowledge tool on their own machine: notes, files, links, auto-organized into a living map. Valuable even if they never sell a thing, which is exactly why they fill it. The filled notebook is the sellable asset.
Paid = fuel
Earnings compound the supply
Money lands → they add more to earn more → a bigger notebook matches more demand → more sales. More experts' notebooks also cross-reference each other, so each becomes likelier to match a buyer. The supply compounds on itself.

Where it does not get double-counted (so this stays honest to a skeptic): when an expert shares their map it rides the same billboard mechanic as Loop 2 (badge embedded, referral credit), we don't count that twice. And the advisor (Loop 8) coaches people to fill it, but the coaching is Loop 8; the supply engine is Loop 9. The one genuinely new thing here is proactive supply: knowledge banked before anyone asks.

2 sides, 1 product

Hobbyists & experts climb a ladder on the left; buyers pull verified answers on the right. The growth job is connecting them.

Supply: hobbyists & experts

Reputation you own · keep 85% of what you charge · founding era: 0% fee
Hobbyist → ContributorSearch, vote, react
SolverResolve clashes,
earn bounties
Senior SolverTrack record,
bigger jobs
Master · LaureateElite, named
in reports
The product 3 AIs clash → a hobbyist or expert resolves it
↑  ↓

Demand: the buyers

Whoever needs a trusted answer
Pay-per-use
Instant Consultant$1.99–$500
one-off
Subscribe
Brands & agencies$99–$499/mo
Enterprise$2.5k → $25k/mo
AI labsdata · by request

A reputation you can't fake

Tie standing to real paid work and you create the credential LinkedIn never could: earned, per-topic, and impossible to buy or buddy-boost.

LinkedIn

Endorsements are people vouching for each other.

Self-asserted and buddy-boosted: anyone can claim anything, and a click costs nothing.
Really Solved

Proof a real customer paid you, and was happy.

Earned per-topic from work someone paid for. Can't be farmed, bought, or vouched into existence.

Everyone who builds standing here walks away with a portable, un-fakeable credential, and a reason to bring their whole network with them.

Hobbyists & experts sell what AI can't

Better models don't shrink the market for human contributors: they clarify it. Hobbyists & experts here aren't fixing AI's mistakes; they're providing 3 things AI is structurally unable to offer, which is why the value is durable at high stakes even as models improve.

✍️
Accountability

An AI can't sign a specific answer and stand behind it. A hobbyist or expert can, and at high stakes, that named accountability is exactly what buyers need. We surface it through attribution; we don't author it ourselves.

Freshness

AI models are trained on the past. A human, hobbyist or expert, is operating right now: they were on-site, in the room, at the conference this week. For breaking events or recent decisions, there is no substitute for someone who was actually there.

🔐
Proprietary context

Their lived experience, unpublished data, and hard-won judgment were never scraped. It exists nowhere on the internet. That un-scrapeable residue is the supply AI can never produce, and the premium the marketplace captures.

This is why we position hobbyists & experts as durable at high stakes, not as AI repairmen who'll be obsolete with the next model update.

The acquisition engine

The viral game portfolio

Each game feeds a different part of the loop: awareness, supply, or demand.

Stump-the-AIs
Awareness

3 AIs clash: the spectacle that pulls people in.

Showdowns
Supply + Demand

2 contributors argue a claim; the crowd votes.

Leaderboards
Supply

Status & FOMO that climbs the contributor ladder.

Catch the AI
Covert expert audition

Spot a model confidently wrong, and prove you know the right answer. Every winner is a latent hobbyist or expert.

Embeddable badges
Awareness

Hobbyists & experts post proof: each one is a backlink.

Quick-Claim Verdict
Demand

Only $0.99 for an instant answer: already written. Hobbyists & experts pre-set their knowledge as sellable, so no one waits on a human.

Creator awards
Awareness

AI-verified "top voice" badges creators want to share.

Referral chain
Viral loop

Bring a friend, both climb: the k-factor engine.

🎯 The "Beat the AI" pattern is a covert expert audition. Every challenge that requires a user to out-know the AI to win is simultaneously recruiting them as a hobbyist or expert: without ever calling it an application. The supply funnel runs through the game layer. Beating every AI (Loop 7: Inbound hobbyist/expert trigger) is the loudest signal of all.
1 funnel, 2 doors

The broad-answer door

Not everyone wants to play the disagreement game. Some just want the broadest, most confident answer fast. We serve both, and the 2 doors are complementary, not competing.

Door 1: the game

The disagreement machine

3 AIs clash. You vote, dispute, escalate to a human. The loop self-recruits, self-monetizes, and generates the verified corpus. This is the supply-and-demand engine: viral by design.

Door 2: the broad answer

Many AIs, one fast answer

For users who don't want to play: aggregate a wide set of AI models and surface the most agreed-upon answer quickly and cheaply. More models raise the floor on cheap/fast answers; humans own the ceiling on high-stakes ones. Together they handle every form of AI substitution from both ends.

Both doors share the same acquisition surface and the same agreement-score infrastructure. A door-2 user can enter the game at any moment: the disagreement score is always there waiting.

The consumer funnel

Free usage manufactures the moment of doubt; the disagreement converts it to revenue; sharing refills the top. (Conversion figures are targets.)

Free searches via AImultisearch or FixTruth100%
2 acquisition doors into the hub: 5/day anon, 15/day registered
Hits a disagreement~20%
≈1 in 5 searches surfaces a model clash: built-in, renewable raw material
Quick-Claim Verdict$0.99
Instant: the answer is drawn from knowledge hobbyists & experts already pre-set as sellable, so no human has to be online
Instant Consultant$1.99–$500
Bring in a hobbyist or expert for a one-off: pay only for what you need, no subscription
Shares & refers: back to the hubk > 0.3
Badges + referral loop feed new searches back into the top of the funnel
🗺️ Funnel → hub model: AImultisearch.ai and FixTruth.com are the acquisition doors: free, viral, pulling traffic from anywhere. ReallySolved.com is the hub where hobbyists & experts answer, earn, and build standing. 1 funnel, 2 doors in, 1 destination.

The buyer landscape

Same verified-truth asset, sold to 6 segments: from a $1.99 one-off to six-figure contracts. Each has a distinct motion.

Instant Consultant
$1.99–$500 · one-off
Consumers pay for a single hobbyist/expert answer on demand, no subscription.
Product-led · pay-per-use
Brands & agencies
$99–$499/mo
Monitor how AI talks about their brand.
Self-serve · email
Market research firms
$20k–$200k / yr
An expert panel cheaper than Gartner.
LinkedIn · demo · pilot
PE / Hedge funds
$50k–$500k / yr
Know the truth before they invest.
Warm intro · 3–9 mo cycle
Brand safety / PR
$5k–$50k / yr
Know before it trends: crisis-driven.
Crisis-moment · fast close
AI labs
pricing on request
Verified accuracy data: cheaper than today.
Hand-sold · the big one
The supply-side secret weapon

"Your interests trending now"

Hobbyists & experts shouldn't have to guess where to show up: this feed tells them. It shows, in real time, what the world is searching for and not getting good answers to. They write the answer and own that knowledge permanently. This is what seeds supply without a cold-start problem.

Feed 1: consumer unmet demand

What searchers can't get answered

Aggregated, anonymized search clusters where AI confidence is low or models disagree. Hobbyists & experts see a topic label and a bucket count ("dozens of searches this week"): never individual queries or identities. Write the answer that owns the demand permanently.

Feed 2: expert-peer demand

What other experts in your field are asking

When a domain expert searches, that's a leading-edge signal: they're at the frontier of the field. Aggregated peer-search demand (same anonymization, verified-expert-only audience) shows what the most knowledgeable people are looking for next. The next-generation signal.

🌐 "Consumer trending now" (next): the same engine pointed at everyone: a public feed of what the world is asking right now, across all topics. This is likely the bigger viral surface (pure curiosity: "what is everyone wondering?"). It ships after the expert feed only because the expert version reuses the demand-clustering backend that's already built and runs on a smaller, controlled audience: the consumer-wide rollout is the next expansion, not a harder one. Aggregate-only, same anonymization rules.
🛡️ Won't top experts hide their searches?: 5 reasons they keep searching here

Your own searches never show up: a topic only appears after lots of different people search it, it's never tied to your name, it's shown broad (a field warming up, not your exact question), it shows up on a delay of about a week, and you're helping fix AI for everyone while you do it. If you still really want to, you can take a few extra steps to keep one single search out of the trend, but by then most people won't bother, because it's anonymous, delayed, and never yours alone. The peer signal shows where the field is moving, never who moved first.

The revenue stack

From quick cash today to the big long-term engine. Same verified-truth asset, sold 4 ways.

Quick cash
Instant Consultant one-offs ($1.99–$500) + $0.99 verdicts + discounted AI subscriptionsImpulse revenue from day one: the on-ramp
live
$9–$25k/mo
Subscriptions: prosumer & company monitoringFrom $9/mo prosumer up to $25k/mo enterprise
live
Easy $
Hobbyists & experts keep the money they'd never have made otherwiseWe capture & organize granular know-how they've been giving away for free: companies worldwide discover their anonymized skills and buy it at the price they set. Founding era: 0% rake. A modest fee phases in as the marketplace matures.
building
The big one
AI-lab accuracy dataCheaper than how labs buy accuracy data today, and the deepest pocket
building

The marketplace fee is the on-ramp, not the prize. The real money is the verified-truth data underneath it. That's why we're charging no platform fee, at least in the first phase, and letting hobbyists & experts keep everything they earn.

Consumer subscription unit economics

$5 / $9 / $19 plans, stress-tested assuming every engine has to route through OpenRouter (worst-case vendor cost) and the fair-use cap we already advertise gets enforced. Max ($29/$99) excluded: shape TBD, built from user feedback post-launch.

$5 Certainizer™ $9 Plus $19 Pro
Blended gross margin/mo~93%~92%~96%
Blended profit/mo$4.67$8.30$18.30
12-month LTV (illustrative)$56.08$99.60$219.60
Compute CAC* (@10% conversion)$0.17$0.17$0.17
LTV : CAC~330 : 1~586 : 1~1,292 : 1
CAC payback period~1 day<1 day<1 day

*We are assuming no ad acquisition spend because of outreach strategies to experts, influencers, & B2B customers, plus all the GTM tools like Clay, Apollo, etc. & the 9 viral loops built into the site. In addition, abusers who create a new email every week will cost us ~$0.80/month.

No one else combines all 5

Each element below is a real standalone product. Combining all 5 into one compounding loop is where the moat comes from.

1Free multi-AI searchThe acquisition engine: pulls anyone with a question
2Viral disagreement gameThe self-recruiting loop: turns searchers into hobbyists & experts
3Verified credentialed marketplaceThe monetization engine: pays hobbyists & experts, wired directly to paying companies
4Auto-generated demand signal"Your interests trending now": everyone's searches become a live map of what to answer next
5Disagreement data for AI labsThe flywheel exit: the corpus grows more valuable with every clash

This combination, confirmed by independent outside analysis, doesn't currently exist as a single product anywhere. That's the category-coinable moment: we can name what we are and own it before anyone else does.

The metrics that matter

A handful of numbers tell us the machine is compounding. (Month-3 targets.)

k > 0.3
Referral coefficient: the loop self-sustains
~20%
Searches that surface a disagreement (built-in supply)
50+
Active Solvers by month 3
500+
Verdicts sold by month 3
$2k
B2B MRR by month 3
65×
Expert CAC payback (the cheap, viral side)
85%
Of every bounty goes straight to the hobbyist or expert
10–100×
Future queries one resolution can answer

The growth role owns these numbers: instrumenting them, then moving them.

See it live: every mechanic is already running

The machine, in the real product

Every feature above is live right now. Click straight through to the exact page where each one runs, not a demo, the actual product. (The live links are the source of truth: pages keep shipping, so links never go stale the way a static screenshot would. Annotated screenshots are being layered on top of these.)

🔍 AImultisearch: the disagreement surface (door 1)

3 AIs answer in parallel; the disagreement score, the 99¢ Quick-Claim Verdict, auto-routing to the 3 best AIs per topic, and the all-👎 capture (Loop 7) all live here.

✅ FixTruth: the citizen-reporter door (door 2)

The "is this real?" surface: humans who were physically present report what AI structurally can't see. The second acquisition door into the hub.

🏅 ReallySolved: the contributor hub

The marketplace: the contributor ladder & reputation, what hobbyists & experts sell that AI can't, earning (SOLVE · SELL · SHARE), demand bounties, embeddable badges, and the buyer landscape.

📡 "Your interests trending now" + the data product

The contributor-facing demand feed (consumer unmet demand + expert-peer demand), and certainize.ai: the verified-truth data & API the labs buy.

What's under the hood 🔒

The mechanics that make it defensible are trade secrets / patent-pending: happy to go deeper on request.

🔒 Expert-matching & routing
🔒 Reputation scoring & anti-gaming
🔒 Multi-path query intake
🔒 Staged inference pipeline
🔒 AI-agreement scoring & escalation thresholds
🔒 Engine-level unit economics
🔒 Lab-data pricing & deal structure
🔒 Patent specifics
Where you come in

Own the machine, not just the outreach

The machine is already built: the product, the loops, the 7 connections, the supply forming. What's missing is the person who makes it fly: someone to pull every ounce of growth out of it, end-to-end, from the funnels to the revenue. Not building from scratch, accelerating what's already running.

DEEPER DETAIL 👇

That's the whole picture: you can stop here. Below is the supply ladder, the go-to-market sequence, the 90-day plan, how the data becomes the prize, and the full operator stack.

The contributor ladder

Standing is earned by doing the work. Each rung unlocks bigger, higher-value jobs, and real income.

Hobbyist → ContributorSearch, vote, react, first verified contribution
status & access
SolverResolve real AI disagreements for pay
$500–$2k / yr
Senior SolverTrack record unlocks bigger, higher-value jobs
$5k–$15k / yr
Master SolverElite tier; sustained excellence across a domain
$50k–$200k+ / yr
LaureateNamed in reports; a press-grade credential
variable
Cold-start solve: we seed the top of the ladder with 50–100 Founding Solvers, pedigreed and invite-only, so the marketplace has credible supply on day one. Disagreement (the demand trigger) is already free and built into every search, so we never have to manufacture it.

How the motions orchestrate

They hand off in sequence: cold-start supply first, then demand, then amplify.

01Founding Solvers

Invite pedigreed experts to seed supply & credibility.

02Consumer launch

Stump-the-AIs + referral loop drive the crowd.

03Buyer outbound

PE/HF, market research, brand safety, compliance.

04Content + social

Substack + automated social amplify the moments.

05AI labs

Sell the accuracy data: warm, and the biggest.

The first 90 days

How it sequences

Supply first, then the loop, then demand and revenue: each phase unlocks the next.

Weeks 1–4

Seed the supply

  • Recruit 50–100 Founding Solvers
  • Stand up bounties + badges
  • Instrument the whole funnel
Weeks 5–8

Launch the loop

  • Stump-the-AIs + referral engine live
  • Tune the share loop toward k > 0.3
  • First verdicts & one-offs converting
Weeks 9–12

Turn on demand

  • Buyer outbound: brands, research, funds
  • Content + social amplification
  • First B2B revenue + lab-data conversations
Why the data is the prize

1 asset, 5 products

Every resolved disagreement becomes a row of ground truth. The same corpus sells to the labs 5 different ways.

1 verified-disagreement corpusevery resolved clash = question × each model's answer × who was actually right
↓   sold to the labs as 5 products  ·  pricing on request
Living eval benchmarksa rolling private test set of frontier-hard cases
Preference pairs for trainingcorrect vs. confidently-wrong, pre-filtered
Error-mode mining"Model X is wrong N% on topic Y"
Hallucination labelsasserted-but-false, verified instances
Reward-model datatraining signal for verifiers

This is why labs are the deepest pocket, and why we're far cheaper than how they source accuracy data today.

A proposed starting stack: swap in your own

Not a mandate: a starting point. Swap any piece for the tools you already trust; this just lays out the moving parts the viral features need wired together. Items marked [NEW] are what the 2026 viral-features plan adds.

ACQUISITION / SIGNALS          ENRICH + SCORE             SEQUENCE + SEND          TRACK
─────────────────────          ──────────────             ───────────────          ─────
AImultisearch · FixTruth ──►   Clay (enrich, score,   ►  Resend (email)       ►   Airtable (CRM)
site visitors / waitlist       dedupe, AI research)       LinkedIn cadences         + growth dashboard
Apollo / RB2B / lists          Blitz (expert lead-gen     Postiz (social posts      attribution by
games arena events [NEW]       → Clay → sequence) [NEW]   + drama-card clips) [NEW] sourced-revenue tag
Inbound hobbyist/expert trigger [NEW]
                                        │
                                        ▼
                           n8n (the glue: triggers, routing, webhooks, retries)
                           ├─ expert-capture routing  ←  all-👎 → inbound hobbyist/expert trigger  [NEW]
                           ├─ demand-radar clustering  ←  unmet search signal → expert match [NEW]
                           └─ viral clip pipeline      ←  drama cards, share-card generation [NEW]
                                        │
                                        ▼
                           Supabase (our data)  ·  trigger.dev (code-first jobs)
                           ├─ search_history · page_feedback · gripes
                           ├─ expert_credentials · notebook_entries (4-tier privacy) [NEW]
                           ├─ candidate_brief_events · bounty_pledges               [NEW]
                           ├─ demand_clusters · expert_radar_feed                   [NEW]
                           └─ advisor_state · support-chat "advisor mode"           [NEW]
Clay: outbound brain n8n: the glue (self-hosted, replaces Zapier) Resend: email Airtable: CRM Apollo / RB2B: signals Supabase: our data trigger.dev: code-first jobs Postiz: social Blitz: expert lead-gen [NEW] Drama-card clip tooling [NEW] Inbound hobbyist/expert trigger pipeline [NEW] Demand-radar clustering [NEW] Viral share-card generator [NEW] Cross-site AI advisor [NEW]

Clay = outbound brain (find → enrich → score → research). n8n = the glue replacing Zapier (self-hosted on OCI). Blitz → Clay is the external expert-recruitment path (own session). Drama-card clip tooling covers the social-share pipeline for viral AI-disagreement content. The inbound hobbyist/expert trigger pipeline and demand-radar clustering are new n8n workflows triggered by existing Supabase signals.

Please let me know if anything is missing or wrong on this page by clicking here 🙏