A visitor lands on your page. Instead of reading a pitch written for everybody, they answer three or four questions in a 90-second chat. While they're still typing, the page is rebuilding itself around what they just said — their words in the headline, their situation in the proof, their goal in the offer, and only the objections they actually raised.
It sits on pages you already have. Nothing new to build, nothing new to maintain, no new URLs in your email sequence.
Three or four questions, then it's over. No persona, no name, no "hi I'm Ava 👋". It asks, then it gets out of the way.
The AI didn't write the page. It decided which of a copywriter's words this particular person sees. That distinction is the whole product.
We checked about sixty products across six categories before building. The market has solved every individual piece of this — and nobody has wired the input to the output.
| Architecture | What it reads | What it produces | Who |
|---|---|---|---|
| Asset factories | A CSV or CRM list | Pages pre-rendered in batch | Hyperise, Sendspark, Potion |
| Runtime selectors | Traffic source, UTM, reverse-IP | Picks between variants a human wrote | Unbounce, Webflow Optimize, Coframe |
| Runtime generators | Session and referrer signals | An LLM rewrites copy per visitor | Fibr AI ($479/mo), Wix |
| Conversation → structure | What the person just typed | Re-composes which blocks appear | Empty. That's us. |
Putting someone's words in the headline. Instapage has done keyword-echo for a decade; Fibr and Wix generate copy per visitor today. Assume any competitor ships this within one launch cycle.
Changing which components exist on the page and in what order. Swapping the bonus stack by stated goal. Building the FAQ from only the objections raised. Every runtime tool in the market rewrites copy inside a fixed layout. None re-composes the layout from what a person said.
If all three hold, AI Mirror works. If any fails, it doesn't — and we'll tell your buyers that in the copy.
Each layer reads a different signal and changes a different thing. They run separately, so a weak answer on one doesn't degrade the others.
Early designs had an Emotional mirror that matched page tone to inferred emotional state. We cut it. Inferring emotion from a free-text "what's your biggest problem" box means storing sensitive data across any real audience — and it was the layer visitors noticed least. Register matching inside the Linguistic layer gives the same visible effect with none of the exposure.
Price mirroring. Offer composition changes; the numbers do not. Every price, guarantee and bonus term is pulled from an approved library the AI cannot edit. Your buyers can put that in writing to their own clients — and for anyone selling to regulated local businesses, that's a feature they'll pay for.
Written so you can lift rows straight into your review, your bonus page or your emails. Every example is real output from the working demo.
| Feature | Benefit | Worked example |
|---|---|---|
| Linguistic Mirror headline |
The first line already sounds like the sentence in their head — so they don't have to translate your pitch into their problem. | Affiliate says: "I'm still building every report by hand." Headline becomes: "Stop building every report by hand — without adding a person." |
| Proof Mirror selected |
They see the one case study that matches their situation instead of scrolling past nine that don't. Relevance beats volume every time. | Practice owner mentions six failed months of paid social. From 34 case studies the page shows the one about reducing enquiry volume: "We were getting 60 and booking 4. Now we get 22 and book 14." |
| Goal Mirror offer moves |
The offer reshapes around what they said they want. Nobody in this market changes what is being sold from a conversation. | Same product, 12 bonuses available. Agency owner sees Implementation Sprint + White-label Report Pack + Retainer Pitch Script. Complete beginner sees First-Client Playbook + Outreach Scripts + Free-Audit Kit. The other nine never render. |
| Objection Mirror assembles |
Handles the one thing stopping them — and, just as importantly, doesn't plant four objections they never had. | Buyer says price is fine, they just don't want another dashboard to learn. Page shows 1 of 14 handlers. Price, refunds, contracts and support are all suppressed. |
| Section omission removes |
Shorter, sharper pages. Irrelevant sections are deleted, not collapsed — which is the thing a static page physically cannot do. | Homeowner says "no insurance, this is out of pocket." The entire insurance-claim section is removed from the quote rather than shown and ignored. |
| Progressive assembly the demo |
The page visibly rebuilds while they're still answering. This is the 15-second clip — the reason people forward it. | Sub-1.8-second visible update after every answer. Sections fill one at a time, on screen, during the conversation. |
| Multi-input no adoption cost |
Chat, in-browser voice, or an imported call transcript. The transcript route means value on day one with no widget on any page. | A consultant uploads a 9-minute Zoom discovery recording. A full proposal, a seller brief and a follow-up email come back. Nobody typed anything. |
| Mirror Seller Brief second output |
One conversation produces the prospect's page and the seller's intelligence: score, needs, objections, decision criteria, recommended next action, CRM fields. | "Score 84. Ready to sign against a fixed scope. Do NOT ask for a second call — she has told you she wants terms, not more conversation." |
| Follow-Up Mirror continues |
Follow-up that resumes the actual conversation instead of "just checking in" — and picks the type of follow-up, not just the words. | Sceptical buyer, no purchase. Engine suppresses the scarcity template and the bonus-stack reminder, and sends candour instead: "What this doesn't do." |
| Mirror Intelligence compounds |
Every conversation is market research. Top objections this week, goal distribution, the phrases your market actually uses. It accumulates — which is why buyers stay. | After 400 conversations: "62% of your traffic names setup time as the blocker, not price. Your sales page leads with price." |
| Approved-library lock deterministic |
Pricing, guarantees, availability and compliance-sensitive claims cannot be generated. The AI selects; it never invents. | Every price block on the page carries a lock line: "Rates come from the published price list. AI Mirror chooses which pieces to propose; it cannot change what any of them cost." |
| Four-tier fallback never breaks |
The page always renders. Worst case it's the normal page — so there is no scenario where this makes things worse. | Junk, abuse or a prompt-injection attempt drops instantly to the customer's ordinary page with zero visitor text echoed anywhere. |
A personalised page is what sells this in fifteen seconds. But the same conversation, read once, produces every asset the next step needs — and that's what makes buyers keep paying after launch week.
A page personaliser is a $47 one-time purchase. A proposal-and-quote engine is a monthly bill a local business owner pays without blinking. Same code. The front end sells on the page because the page demos in fifteen seconds — and the back end is where your buyers keep getting value, which is what protects your refund rate.
The Seller Brief is the honest safety net. Even if personalisation only lifts conversions modestly on their traffic, they still get a structured intelligence report on every lead — which is worth the subscription on its own. Very few products in this space have a second reason to exist.
This is the comparison to put in your review. Not a before-and-after against a competitor — a side-by-side of the same product assembling differently for two people.
Note what happens on the right: the engine talks a sceptical buyer down. That's the "Should I buy?" behaviour — and it's the single most trust-building thing in the demo. Affiliates who lead with it get a different quality of reply from their list.
| What they said | Visitor A — hail damage, active leak | Visitor B — 22-year roof, selling in spring |
|---|---|---|
| Recommended step | 20-minute leak & hail inspection. Replacement deliberately not offered — they said they feared being upsold. | Written pre-sale condition report with costs itemised by line. |
| Comparison table | Ranked by speed: water stopped same visit vs 2–3 weeks. | Ranked by cost, with a resale-recovery column added. |
| Insurance section | Claim checklist + filing-window warning. | Section removed entirely — no insurable event. |
| Follow-up if unbooked | SMS: repair-first promise in writing + the real filing deadline. | SMS: how to read any roofing quote. No urgency, no discount — the sale is months away. |
The florist build is worth showing even to buyers who will never sell flowers, because of one section: the Seasonal Note. It tells the couple what isn't available on their date.
Dahlias, cosmos and amaranthus at peak. And in the same block, in writing: "Peonies — June only." If they'd pinned a photo of peonies, the page says so now rather than in August.
Amaryllis and ranunculus recommended. And: "almost nothing is British-grown, and your date sits inside the Valentine's window when wholesale prices rise sharply." The uplift is priced in rather than added later.
Two matrices. First the four layers — how the mechanism shows up for each audience. Then the nine modes — what each audience actually gets out of it. Use this to pick which examples go in your promo, because the wrong example for the wrong list kills an otherwise good angle.
Sells services to local businesses. Lives on discovery calls, proposals and retainers. Highest willingness to pay, longest sales cycle, and the buyer who keeps paying monthly.
Promotes other people's launches. Lives on bonus pages, review pages and a list. Buys fast, refunds fast, and is competing against 400 identical bonus pages.
Sells their own course, membership or coaching. Lives on webinars, VSLs and applications. Their real enemy isn't conversion — it's refunds and non-completion.
Same mechanism, three very different-looking results.
| Audience | What the prospect said | What the page opens with |
|---|---|---|
| Agency | "We get calls, but they're all tyre kickers." | "You're not short of enquiries. You're short of the right ones." — and the whole proposal is framed as reducing volume, which is the opposite of what every other agency pitches. |
| Affiliate | "I'm still building every report by hand." | "Stop building every report by hand — without adding a person." The product's actual feature list never appears above the fold. |
| Creator | "I've bought three courses and finished none of them." | "You don't need another course. You need to finish one." — then the page leads on completion mechanics rather than curriculum. |
Note for the creator use case: that headline is the single highest-leverage line in the whole use-case book. Course buyers' dominant private objection is their own history of not finishing, and almost no sales page addresses it because addressing it feels like introducing doubt. Mirroring lets you say it only to the people who said it first.
| Audience | Signal read | What gets shown — and what gets hidden |
|---|---|---|
| Agency | Industry, revenue band, prior failed channel | A dentist sees the dental case at a matching practice size. From 34 case studies, one renders. The seven-figure client logos are deliberately suppressed — they make a $40k/month practice feel like the wrong customer. |
| Affiliate | Business stage, client count, list size | A five-client operator sees the testimonial from a four-client operator — not the guru who did $2M. Aspirational proof under-converts practical buyers; it reads as "this isn't for me." |
| Creator | Prior identity, background, constraint | A nurse retraining into UX sees the student who was also a nurse. Course proof is identity proof: the buyer isn't asking "does this work", they're asking "does this work for someone like me". |
| Audience | Two goals, same seller | Two different offers |
|---|---|---|
| Agency | "Ten implant consults a month" vs "Keep three chairs full" | Implant Case Acquisition System vs Chair Utilisation System. Different deliverables, different reporting, different success metric — from one sentence. |
| Affiliate | "Add three retainer clients" vs "Land my first client" | Implementation Sprint + White-label Reports + Retainer Pitch Script vs First-Client Playbook + Outreach Scripts + Free-Audit Kit. Three of twelve bonuses render; nine never appear. |
| Creator | "Get my first paying client" vs "Scale past $10k months" | Curriculum re-orders — Module 4 is foregrounded for one, Module 9 for the other. The default tier flips from self-study to cohort. The bonus swaps from templates to a peer group. |
| Audience | The objection they actually raised | What renders — and what is suppressed |
|---|---|---|
| Agency | "I've been burned on retainers." | Fixed-scope clause, dates against every deliverable, and a miss-the-target-and-month-three-isn't-invoiced line. Price, contract length and team-credentials handlers all suppressed. |
| Affiliate | "I don't want another dashboard to learn." | The setup-call handler, alone. 1 of 14. Price, refunds, contracts and support never render — she told you price was fine, so raising it would create a doubt that didn't exist. |
| Creator | "I never finish anything I buy." | Completion mechanics: fixed cohort dates, 20-minute lessons, an accountability partner, and the actual completion rate. The "is it beginner-friendly" and "lifetime access" handlers are suppressed — lifetime access is a negative to this buyer. |
That last cell is the clearest demonstration of the layer in the whole document: for most course buyers "lifetime access" is a benefit. For the buyer who just told you they never finish things, it's a permission slip to procrastinate. A static page cannot make that call. This one does.
Priority grid first, so you know which examples to lead with for which list.
| Mode | What it produces | Agency | Affiliate | Creator |
|---|---|---|---|---|
| Mirror Sell | Personalised page | ●●● | ●●● | ●●● |
| Mirror Propose | Proposal / scope | ●●● | ● | ●● |
| Mirror Quote | Estimate / configuration | ●●● | ●● | ●● |
| Mirror Recommend | What you should choose | ●● | ●●● | ●●● |
| Mirror Compare | Weighted comparison | ●● | ●●● | ●●● |
| Mirror Report | Audit / diagnostic | ●●● | ●● | ●●● |
| Mirror Plan | Roadmap / study plan | ●●● | ●● | ●●● |
| Mirror Follow | Continuation, any format | ●●● | ●●● | ●●● |
| Mirror Book | Booking experience | ●●● | ● | ●● |
The Mirror Seller Brief isn't a tenth mode — it's generated alongside every one of the nine. Agency: a lead dossier before the callback. Affiliate: segment and objection data across the whole promo. Creator: an applicant brief before an enrolment call.
On their own site, after an audit-request opt-in. Right now that site shows a plumber the same page it shows a dentist. After: industry case study, matching revenue band, the service package for the gap they named, and the handler for their specific agency trauma — "the last one ghosted me" is a different page from "too expensive".
The bonus page, post-opt-in. Every affiliate promoting the launch has the same stack. This is the only differentiator that isn't "add a 21st bonus". Also works on the review page after the video, where intent is highest and almost nobody does anything with it.
The webinar replay page and the VSL page. Replay traffic is the biggest wasted asset in info products — same replay to a hot buyer and a tyre-kicker. After: high-intent sees offer and buy; mid sees comparison, FAQ and proof; low sees education and a reminder.
Upload the discovery recording, get back a proposal with their situation in their words, the matched case study, a scoped deliverable list, dates and the approved price. The Tuesday call that's still unwritten on Friday becomes a Tuesday-afternoon send. This alone justifies the agency tier.
Honestly, the weakest cell in the grid — pure affiliates don't write proposals. It becomes relevant the moment they sell anything done-with-you off the back of a promo, which most eventually do. Don't lead a purely affiliate promo with this mode.
High-ticket coaching and masterminds. The application becomes a conversation, and the applicant receives a personalised enrolment proposal — which programme track, which student result matches them, which payment option is defaulted from the published set.
Custom scope priced from the conversation, itemised by line, against the published rate card. Competing on legibility rather than on being cheapest — which is how you win against a lower quote that nobody can read.
The honest total-cost quote: "based on what you told me, here's the front end plus the two upgrades you'll actually use — and the three you can skip." Counter-intuitive, and it converts, because everyone else on that launch is selling the $900 full stack to a beginner who needs the $47 one.
Which of the published payment options is presented first, and what the total looks like across the three tiers. The prices are fixed and identical for everyone — what changes is which one leads.
Which service package fits, with the reasoning shown. Also the version that says "you don't need us yet, fix your Google listing first" — which loses a $2k month and wins a referral network.
"Based on what you told me, here's whether this is right for you — and where it isn't." Includes recommending not buying when the fit is poor. It costs a percentage of sales and buys a list that opens the next twelve emails.
Catalogue triage. A creator with nine products loses buyers to choice paralysis at the "start here" page. This turns it into: "start with this one, skip these two, come back for that one in about six months."
In-house hire vs freelancer vs agency, with the columns ranked by whatever they said matters — speed, cost or certainty. The comparison a prospect is running in their head anyway, done in the open.
Tool A vs Tool B, weighted to their workflow. The comparison post is the highest-converting affiliate format there is, and it's always written for one imagined reader. This writes it for the actual one.
Your course vs the free YouTube route vs the $47 competitor — argued honestly against their stated constraint. Time-poor buyers should be told the free route costs them 60 hours. Cash-poor buyers should be told to take it.
The free audit, generated from a two-minute conversation instead of four hours of manual work. This is the classic local-consultant opener, and the reason most consultants stop offering it is that it doesn't scale. Now it does.
A readiness score before purchase: "here's what you'd need in place for this to work, and you have two of the four." Pre-qualifies buyers, which is the cheapest refund reduction available to an affiliate.
The skills-gap diagnostic as the lead magnet. Answer six questions, get a report naming your three gaps — and the paid product is the thing that closes them. The report is the sales argument, made about them.
The 30/60/90 built backwards from their date, not forwards from signature. When a prospect names a deadline — "we open the second location in March" — the plan states the last safe start date. The deadline closes the deal for you.
A 14-day implementation plan delivered with the purchase, built from what they said they wanted. Refunds in this market are mostly non-use, and non-use is mostly "I didn't know where to start." This is the single highest-ROI mode for an affiliate's refund rate.
A study plan sized to the hours they said they have. Someone with four hours a week gets a twelve-week path; someone with fifteen gets a five-week one. Same course, two completion rates.
Proposal opened but unsigned → the follow-up addresses the section they lingered on and the objection they raised on the call. Not "just circling back". And it can choose to send a revised proposal instead of an email.
Didn't buy → the sequence branches on why. Setup-friction buyers get the four-minute configuration recording. Sceptics get candour. Nobody gets the cart-closing email if cart-closing wasn't their problem.
Before purchase: the specific blocker, addressed. After purchase: onboarding follow-up that references what they said they wanted, aimed at getting them to lesson three — which is the point past which almost nobody refunds.
The booking page shows the right call type, the right length, what to bring, and a pre-brief that lands in the consultant's inbox before the call. Fewer no-shows, and no first ten minutes spent on discovery you already have.
Mostly not relevant — no calendar in the funnel. The exception is high-ticket promos with a bonus onboarding call, where booking rate is the whole commission.
Cohort selection by stated availability, application calls routed by fit, and office-hours slots surfaced by the topic they said they were stuck on.
Every page renders at one of four quality tiers, chosen by an extraction-confidence score. The page always renders. Worst case, it's the page they already had.
Transparency is also the higher-converting design. The research on personalisation is consistent on this: covert personalisation reduces response because it triggers privacy concern, while overt personalisation increases it. "Here's why you're seeing this page" is a conversion asset, not a compliance tax.
| Tier | Price | What's in it |
|---|---|---|
| Front end | $47 | Engine + 5 done-for-you master templates + 3 campaigns + 1,000 conversations/month |
| OTO 1 | $97 | Mirror Vault — 30 professionally written master templates. Highest take-rate tier; the templates are what make the magic work. |
| OTO 2 | $147 | Pro — unlimited campaigns and conversations, in-browser voice, multilingual |
| OTO 3 | $97 | Mirror Intelligence — aggregate voice-of-customer analytics across all their traffic |
| OTO 4 | $197 | Agency — 20 sub-accounts, commercial rights, white-label |
| OTO 5 | $67 | Live training — template authoring and personalisation mastery |
| Bundle | $397 | Everything |
$17 buys a tyre-kicker who never logs in and refunds at day 29. $47 buys someone who installs it. Your EPC comes from the OTO take-rate, and the OTO take-rate comes from FE buyer quality.
OTO 1. The Vault is the actual product — a real copywriter writes roughly 580 blocks per pack. It's gated for a reason, and it converts because buyers hit the limits of five templates fast.
You can run this on your own bonus page for this very launch. Every affiliate promoting a launch has the same bonuses as every other affiliate. This is the only way to differentiate a bonus page that isn't "add more bonuses".
Each angle leans on a belief your audience holds before they hear about the product. Don't stack them — pick one, commit to it for the whole promo, and let the demo be the proof beat.
Pure curiosity. No problem framing, no agitation — just the fifteen-second clip and a question. Highest ceiling, works on cold and warm traffic, and it's the one most likely to get forwarded inside affiliate groups.
Lead line: "Watch the right-hand side of this screen. She hasn't finished typing yet."
The sharpest angle for an affiliate list, because it's true and everyone knows it. Four hundred affiliates promote the same launch with roughly the same bonus stack. The only lever anyone pulls is more bonuses. This is a different lever.
Lead line: "There are 400 bonus pages for this launch. 398 of them are the same page with a different name at the top."
Most of your list cannot run a valid A/B test — they don't have the traffic, and they half know it. Reframing personalisation as "optimisation for people without traffic" turns a weakness into the reason to buy.
Lead line: "You need about 2,000 visitors before a split test tells you anything. How many did you get last month?"
Every marketer has agonised over which angle to lead with. This angle says: the argument is unnecessary, because the visitor will tell you in ten seconds if you ask them. Very strong on experienced, copy-literate audiences.
Lead line: "You've spent eleven years learning to guess what your market wants to hear. What if you just asked them, on the page, and the page changed?"
The thank-you page is the highest-intent, lowest-effort real estate in the entire funnel, and almost everyone leaves it as "check your inbox". No new traffic, no new pages, no new spend — just the space they already own.
Lead line: "Go and look at your thank-you page. It says 'check your inbox.' That's the most valuable ninety seconds you own and you've put a full stop in it."
A static FAQ handles the objections you imagined. It also plants four they never had. This angle is quiet and logical — it converts thoughtful buyers rather than excitable ones, and it produces low refunds.
Lead line: "Your FAQ answers seven questions. Your buyer had one. It probably wasn't on the list — and you just introduced them to six new worries."
The strongest angle for a burned list, and the one most likely to get replies rather than clicks. It's counter-intuitive enough to earn attention and true enough to survive scrutiny — the demo really does tell a bad-fit buyer to close the page.
Lead line: "I watched it tell a prospect not to buy. I'll show you the screen recording, because I didn't believe it either."
Everyone in this market sells you more — more bonuses, more sections, more proof, more urgency. This is the only tool whose party trick is removing things. Refreshing, memorable, and structurally impossible for a static page to copy.
Lead line: "Every tool you've bought this year added something to your page. This one deletes — and it deletes the bits your visitor just told it were irrelevant."
The angle for consultants and agency owners, and the one that supports the highest price. Not "a better page" — the whole post-call admin pile, gone. Upload the recording, get the proposal, the brief and the follow-up.
Lead line: "You had the discovery call on Tuesday. It's Friday. The proposal is still not written. Upload the recording."
Every personalisation tool on the market needs to already know who the visitor is — a CRM record, a reverse-IP lookup, a CSV row. A conversation removes that requirement entirely. This lands hard on anyone who has priced Mutiny or Hyperise.
Lead line: "Enterprise personalisation costs $479 a month and needs to know who your visitor is before it can do anything. Ours just asks them."
Colour-coded by the angle they belong to — spectacle & contrarian logic & leverage pain & specificity. Mixing angles inside one promo is the most common way to flatten a good one.
Everything below exists to protect your refund rate and your reputation. A promo built on the mechanism survives day 30. A promo built on a multiplier doesn't.
AI Mirror — Partner Kit. Prices, tiers and asset lists are current as of this document and may change before open cart; nothing here is a contractual commitment. Figures cited for prior launches are historical and are not a projection of your results. No earnings claim is made or implied anywhere in this document.