All eighteen stand between a brief and a payment on Nexertise. An engagement that is rejected is not charged, and the budget unit is refunded.
Each layer is independent, and every one of the eighteen is named below with what it blocks.
Government-issued ID verification (Stripe Identity) before earning. Every account that engages with a paid campaign is tied to a real, named person before any work is possible.
A banned account’s identity and email are kept on file as hashes. A new account whose verified identity matches one is banned at verification.
A persistent standing record follows every account across sessions. Each quality strike cuts the six-month reputation score that sets earning tier and access.
Engagement volume and content access are rate-limited per account. Gating happens before any content is delivered.
Live biometric checkpoints during a session, bound to the device’s hardware authenticator. Biometric data never leaves the device.
Content is delivered in a form that resists automated extraction.
Each delivery is uniquely fingerprinted per account. Any leak is forensically traceable back to its source.
Continuous, foreground attention is required. Inattention, multitasking, and background sessions do not count toward completion.
Response and interaction timing are continuously analyzed. Patterns inconsistent with genuine human attention are flagged in-session.
Account behavior is continuously profiled across sessions. When a verified account is handed off to automation, the drift is flagged in-session.
A comprehension check on the actual content, with a high pass bar. Questions and ordering vary per professional, per session.
Submissions are scanned for signals that the content did not originate with the reviewer. Detection auto-flags the engagement.
The submission is checked for genuine engagement with the specific pitch under review. Generic responses fail.
Multiple authorship signals are evaluated on the submission. A reviewer’s own voice doesn’t trigger; machine-pasted output does.
A dedicated AI-text classifier independently evaluates the submission. Verdicts are logged and audited.
All in-session signals fold into a single per-account assessment. It never rejects a submission on its own: it slows how fast money moves and when an account is asked to re-verify, and a rejected engagement returns the budget unit.
A person reviews every submission against a consistent quality standard before payment is released, assisted by an AI model that does not decide on its own. Professionals can contest a rejection, and members in the EU, EEA and UK can ask a person on our team to review it.
Budget escrow, atomic operations, and automatic refunds. Funds release only after the submission is approved — otherwise the unit is returned to the advertiser. New accounts earn their way to full withdrawals on a graduated schedule, flagged accounts re-verify before cashing out, and payouts track the payment that funded them so card fraud never reaches honest professionals. Withdrawals are paid only to the professional’s verified payout bank account, and when a professional adds or changes their payout bank account, withdrawals may pause for up to 7 days while the new account is confirmed. If an account is closed by a security check, its funds are held while the account is closed, not forfeited, and the owner can appeal.
Every one of the eighteen layers is named above, with what it blocks. Every submission is reviewed before an advertiser is charged, and a rejected engagement is refunded.