Legal · Lens
Lens
The compliance product, in depth: every claim in every creative, checked against the rules that bind you, with the evidence, before it goes out — and for each part of it, plainly, whether it is built or coming.
Last updated · 2026-09-12
What Lens does
Lens is the compliance product: every claim in every creative, checked against the rules that bind you, with the evidence, before it goes out. This page says what that means, part by part, and for each part says plainly whether it is built or planned.
A creative — an image, a video or a document — is read: an image or a video as a whole, with a video's spoken words transcribed; a document from its first page up to a fixed limit of pages, with the check's stored record saying how many of its pages were read. Every claim in what was read is checked against the rules for its market and its category. Each finding cites the clause it rests on and quotes the element it is about, word for word. The result is pass, needs substantiation, or fail: needs substantiation means the finding is real and the evidence that clears it lives outside the creative, and the result names the evidence it needs. For a video or an image, the bytes that were judged are kept, with their fingerprint; a document handed in for a check is judged as it arrives, and only its fingerprint is kept.
Keeping a document's bytes on record the same way is coming.
The check runs at the doors this page names: the studio's post step, the moment a scheduled post fires, the connected channels a brand publishes to, and the brand's assistant. At the studio's step, at the firing of a scheduled post and at the connected channels, while the stop is off, the check writes its finding — passed, or would have been blocked, with the rules it cites — to the brand's event log, and the publish goes through; at the studio's step and at the firing, when a studio creative is attached, an approval door comes first, and a creative that door refuses never reaches the check. A post the brand's assistant is asked to publish or schedule is different: a finding, if there is one, is shown on the card the person confirms, and a failing result refuses the action, with or without the stop. Copy can also be pre-checked as plain text before it is published, in a fast pass that flags a claim to review and never fails a creative on its own.
A hard stop — a failing result refusing the publish — is built into the studio's step, the firing of a scheduled post and the connected channels, and held off by default there; switching it on is coming, and so is putting the finding in front of the person who decides at those doors.
The promise the plan makes for Lens reads wider — every claim checked before it goes out and after — and the check kept up after a creative has gone out is the next step of Lens, continuous brand compliance: it is coming, and it is not what Lens does yet.
See also: Security & data
Rulebooks
A check reads rules from more than one place: the regulator's, the sector's, and the brand's own.
The regulator's rules are held as versioned rule-books, one for each regulator and market we cover, each stamped with a fingerprint of its exact contents, so a result records the exact version it was made under. A mechanical screen reads the creative against them: a disclosure or a disclaimer that must be present, a label that must be shown, a category that may not be advertised at all.
The sector's rules are chosen from the creative's category and read beside the baseline every creative gets, in an order the category sets: for a category whose advertising is restricted or barred outright, the sector's rules are the gate and are read first — before the baseline, and before any question of proof; for every other category the baseline is read first and the sector's rules after it.
The brand's own guideline is the other source: the guardrails written in the brand's brain — what it must not claim without proof, the disclaimers it always carries — are read into the check at publish beside the regulator's rules and the sector's.
What happens on a hit is decided by the rule-book itself, and what it decides is a label: a hit on a rule marked as an absolute bar is marked blocked, a mechanical failure or a claim that needs proof is marked for review, and a creative with neither clears. What refuses on the label today is this: the studio refuses to approve a creative marked blocked; and wherever a studio creative goes out or is handed on — posted now, scheduled, fired at its time, published by the brand's assistant, assigned to someone in the console or delivered on a request — only an approved creative passes, and an approved creative marked blocked or for review is refused as flagged, with or without the hard stop. The hard stop, off by default, adds refusals of the check's own: at the publish doors on the check's result, and at the studio's approve step on the server's own reading of the label.
Handling per rule — block, review, inform, or accepted-risk, chosen by the brand rule by rule — is coming.
Monitor surfaces
Monitoring of the surfaces a brand publishes to — its site, its social accounts, the marketplaces that list it, its partners' pages, its paid placements and its communications — is coming: a creative found there would be read as a check reads a submitted one.
What runs on a schedule today is narrower and points the other way: a registry of the pages the regulations are published on is checked daily, and a fingerprint of each page is written to the run's log.
Nothing compares a fingerprint with the one before it, and no one is told of a change.
The remediation loop
A finding comes with the fix: each result carries recommended actions beside its findings. Inside the console a creative moves from draft to review to approved to published; a reviewer approves it or sends it back, and sending it back requires a reason.
The loop's last step, verified, is coming. Verified means the fixed creative was checked again and the check came back clean: a finding is closed by a clean re-check, which is the proof it was fixed, never by a person's word alone.
Impersonation and counterfeit
Protection against impersonation and counterfeit — a page, an account or a listing that presents itself as your brand, or a product sold as yours that is not — is coming; nothing in Lens finds either yet.
Partner scorecards
A door for an agency is built inside the console and switched off: a brand invites an agency, chooses whether it approves every creative by hand or lets a clean check pass on its own, and every creative the agency submits lands in the brand's review queue with its check beside it, to be approved or sent back with feedback. While the switch is off, that door answers not found to everyone.
Partner scorecards — each agency's or partner's creatives and how their checks went, read over time — are coming.
Reports and exports
Every check produces a report at its own link: the result, each finding with its clause, and the recommended actions. The link opens without an account, so it can be shared, and a link that is wrong answers not found. Inside the console every check a person has run is listed, and only that person can read those rows: the database enforces it.
Exports — a check, or every check over a period, as a file a brand keeps — are coming.
Onboarding
Today one check can be run without an account: on our public check page, anyone uploads a creative, leaves an e-mail address and receives the report's link. An account is by request: the sign-up page takes a request and we review every request. Inside the console a brand's brain — the facts and guardrails the check reads — is built from the brand's own site or from a guided set of questions.
Self-serve onboarding — an account, a brain and a first check without writing to us — is coming.
API and CLI
An API a customer builds on, with its documentation, and a command-line tool are coming, later than the rest of this page. The check's endpoint exists for our own surfaces and is not yet published for customers; the docs page lists no endpoint for it yet.
Security and data handling
What is in place today is written on the security page: workspace isolation at the database level, encryption in transit and at rest, and a brand's own data — its brain — held in India. A check's record is readable only by the person who ran it; the database enforces it row by row.
One engine, two lanes, is the design: an authority's work and a brand's work through the same fetchers, the same evidence store and the same judge, never seeing each other, with an authority's data never visible on the brand's side.
Today, on every table the authority engine writes, a row belongs to one workspace and is visible only inside it, and an authority seat is refused every row of those tables, even when that seat also belongs to the workspace. The database enforces both rules row by row, and both are tested.
A lane written on every row, with a test that a row on each side is invisible to the other, is coming.
ReguNow Surveillance, for an authority that oversees a market, is the authority side of the same engine — a separate product.
See also: Security & data
Request a quote
A quote is by request, and no figure is published: that is the design. There is no list to read here and no form to fill; to ask for a quote for Lens, or to be told when self-serve onboarding opens, write to us at the address our About page publishes, and say which brands and which surfaces you have in mind.
We will answer with what Lens does for your brands as it is built, and say plainly what of this page is still to come.