We paid EUR 30 to become a lead. Then we audited everything. | SerpCtrl
2026-08-10·15min
We paid EUR 30 to become a lead. Then we audited everything.
case study auditAIdigital doublemarketing funnelsAI toolsSharpify
The verdict, precisely
Sharpify is not a fraud in the legal sense. It is a real, registered, tax-paying Latvian company with a real founder - who genuinely is on the Forbes Baltics 30 Under 30 list - and real software.
It is also a textbook low-ticket-tripwire-to-high-ticket-retainer funnel, wrapped in inflated trust metrics, containing two genuinely wrong and one genuinely dangerous technical claim, and carrying a multi-year, cross-rebrand pattern of refund-refusal and high-pressure-sales complaints.
The EUR 30 materials are a competent beginner onboarding. The real product is the retainer quoted on the follow-up call - around EUR 2,800 for three months, in the offer we traced. And the retainer is exactly where the complaints cluster.
Hold both of those at once. That is the whole case.
The 130-slide workshop deck provides beginner-level guidance covering topics such as CLAUDE.md, MCP, a “5 levels” maturity ladder, and 20 different use cases. The original Google Slides and Google Docs sources are linked within the document itself, meaning part of the product consists of shared Google files.
The “Pro plan” resource document contains a collection of approximately 28 AI tools with pricing information, a seven-section CLAUDE.md template, a 10-point business-context questionnaire, and 27 “skills”.
The 27 “skills” are the genuine value in the package. They are competent prompt templates. For example, the offer-creation prompt is based directly on Alex Hormozi’s frameworks (including the Value Equation and concepts such as “price at 10x, sell at 1/3”). The Meta ads prompt provides a real 46-point audit framework, while the cold-email, lead-research, funnel, and onboarding prompts are also solid and practically useful.
The video training is simply provided as a YouTube link.
The free CRM trial gives access to app.sharpify.io, their own SaaS product, which also serves as the entry point for the upsell process.
Then there is the catch, and it is not a small one.
The document instructs you to paste each "skill" into .claude/skills/[name]/SKILL.md and says Claude Code "will automatically find and use it." As shipped, the skill texts begin with ROLE / INPUT / PROCESS and have no YAML frontmatter - no name:, no description:. Without that frontmatter, a skill does not register for automatic discovery. This is not an obscure edge case; it is the documented mechanism the entire feature depends on.
Which means the deck's central promise - copy, paste, and your AI automatically becomes you - is broken as documented. The 27 templates work only as manual prompts you feed to the model by hand. The flagship feature of a product about AI automation does not automate.
The company behind it
The founder is Niks Jansons - real, publicly identifiable, and reputationally exposed. TEDxRTU speaker, covered by Dienas Bizness, Diena, and Forbes Baltics, co-host of the "Stay Sharp Sovs" podcast with fintech entrepreneur Matiss Ansviesulis, headliner of Stay Sharp LIVE, a genuine ~750-attendee 2025 event at ATTA Centre in Riga. None of that is invented. Keep it in the ledger's credit column.
The corporate structure is three entities, and the registry tells a story the marketing does not:
The company structure consists of three entities.
SIA Sharpify Services (registration number 42103096107), previously known as SIA NJ Media until 15.11.2025, is the operating company. It has been VAT-active since 2019, with share capital of EUR 2,800 and 10 employees reported in 2024. The company generated approximately EUR 787,000 in net revenue and paid between EUR 130,000 and EUR 207,000 per year in state-budget taxes during 2022–2024. The company is 100% founder-owned.
SIA Sharpify (registration number 50203132681), previously known as NJ Holding SIA until 2019, appears to be a near-dormant shell company. It has EUR 100 in share capital, reported zero employees during 2022–2024, and was flagged by the State Revenue Service (SRS) on 04.09.2025 as an inactive taxpayer with no declared transactions for more than six months. This entity holds the Sharpify brand and portal.
Sharpify Global SIA (registration number 40203719899) exists as a separate entity, but its exact role has not been confirmed. It is likely intended to serve as the international SaaS vehicle.
Note the renames. NJ Holding became Sharpify in 2019. NJ Media became Sharpify Services in November 2025. A rename is legal and sometimes sensible. But multiple reviewers on Latvian complaint portals explicitly read the second one as detaching the operation from the complaint history attached to "NJ Media" - and when we traced that history below, the complaints do indeed span both names, unbroken.
One more structural note. The SaaS at app.sharpify.io is sometimes assumed to be a white-labeled GoHighLevel install. It is not - fingerprinting shows a custom Next.js application on Vercel. It is, however, a feature-clone of that category: 200M-contact B2B lead database, pipelines, visual workflow builder, invoicing, scheduling, "8,000 integrations", sold as an AppSumo lifetime deal at USD 59/129/199. Real software. Crowded category.
The claims, against the record
Every trust signal in the funnel, checked:
The audit reviewed six major trust and marketing claims.
The claim that Sharpify was featured in Forbes Baltics 30 Under 30 is TRUE. It is corroborated by Forbes Baltics’ own records in the late-2022 “30 lidz 30” edition. However, the distinction is important: this refers to the licensed Baltic list, not the global Forbes 30 Under 30 ranking. The wording describing it as a “cover feature” appears to be self-asserted and is not independently confirmed.
The claim that Sharpify is the “#1 business show in Latvia” is INFLATED. No official ranking organization confirms this position. Apple Podcasts Latvia rankings place it around #4 in the Business category and around #8 in Entrepreneurship. Additionally, it is a personal co-hosted podcast rather than a Sharpify product.
The claim of “2,300+ clients served” is UNVERIFIABLE. This figure is self-reported. Independent GetLatka data shows approximately 25 active customers and an average retention period of around three months. The “2,300+” figure appears to represent a cumulative lifetime customer count, largely influenced by churned low-ticket AppSumo buyers.
The claim of “1,000+ reviews” is REFUTED. Available platforms show approximately 113 reviews on Trustpilot, 2 on AppSumo, and 19 on RatingFacts. Combined, the total remains well below 200 reviews. The company’s homepage itself does not appear to claim 1,000 reviews; instead, it reuses the client count as the denominator for its rating presentation.
The claim that Sharpify has generated “EUR 50M+ in client revenue” is UNVERIFIABLE. It is an unaudited figure attributed to clients, with no publicly provided methodology. The claimed amount is approximately 40 times larger than the company’s own revenue.
The claim of a “15+ expert team” is INFLATED. Registry records show 10 employees in 2024. GetLatka reports a larger figure of 16 people, but this includes a seven-person sales team.
One claim true with an asterisk, five inflated, refuted, or unverifiable. That ratio is the point. A trust stack where the only independently confirmable item is a 2022 list placement is not a trust stack. It is set dressing for a sales call.
The technical audit - including the one that will cost someone their files
The deck teaches Claude Code mechanics, and most of the basics are fine. Three things are not, and because the audience is explicitly non-technical, they matter more here than they would in a developer forum.
First, the dangerous one. The deck claims that pressing Esc twice "undoes EVERYTHING - files, bash, API calls - better than git undo." This is false, and Anthropic's own documentation says so directly: checkpoint rewind tracks only the AI's own file edits and the conversation. It does not undo files changed through shell commands - a deleted file is deleted - and it can never reverse external side effects: API calls made, emails sent, deployments pushed, database writes. The documentation states, verbatim, that it is "not a replacement for version control." The deck teaches the exact inverse. A beginner who believes this reassurance while running destructive commands will lose work and expect to get it back. They will not.
Second, the reckless one. The deck teaches beginners to run the tool with --dangerously-skip-permissions for unattended overnight jobs. The flag does what its name says: it removes every confirmation prompt, allowing any shell command - file deletion, API spend, production pushes - with zero human review. The vendor put "dangerously" in the flag name on purpose. Teaching non-technical marketers to run it overnight, combined with the false "Esc Esc restores everything" promise, is a genuine footgun: permission to do damage, plus a fake safety net.
Third, the stale one. "Fast mode - Opus 4.6 only" was outdated at time of sale. Minor on its own; it dates the deck.
For fairness, the fact-check cut both ways. The deck's showcase projects (Paperclip, Karpathy's autoresearch) are real, not hallucinated. Its McKinsey and Gartner citations are quoted correctly. But its "42,000 GitHub stars in 3 weeks" figure conflates a later cumulative number with the launch window, and its "37% of companies are replacing employees with AI" stat is misattributed to "Gusto Research" - the actual source is a Resume.org survey, it is about expectations for end-2026 rather than the present tense, and Gusto's own published data points the opposite way. A pattern, not an accident: every error inflates urgency.
The funnel is the one they teach
Here is the shape of the offer, reconstructed from the materials and the booking flow:
EUR 30 workshop (ClickFunnels checkout)
|
v
booked video call
|
v
~EUR 2,800 / 3 months "done-for-you marketing + results dashboard" (roughly EUR 933/month, plus their CRM)
The elegant detail: the EUR 30 package includes a "Low-Ticket Sales Funnel" skill - tripwire, order bump, upsell, main offer - which states that the tripwire's job is to "pay for ad spend and qualify buyers." The funnel you experience as a customer is the funnel they sell as a template. You are not buying the map; you are the terrain.
To be precise: a tripwire funnel is a legitimate, widely used sales model. We do not fault the architecture. The questions are only whether the claims that power it are honest (see the table above) and whether the back-end delivers (see the next section).
One documented pattern worth flagging on its own: no retainer price appears anywhere in public. The EUR 2,800 figure surfaces only on the call. Adjacent prices documented in complaints: EUR 3,388 for courses, EUR 5,800 for six months of coaching, EUR 10,000 for done-for-you. Hidden-until-the-call pricing is not illegal. It is, however, the single strongest tell that the price is set by what the call reveals about you, not by a rate card.
The complaint record, both columns
We traced reviews and complaints across Trustpilot, AppSumo, RatingFacts, sudzibas.lv, and navuzticibas.com, 2021 through 2026, under both company names. The record is sharply polarized: the cheap SaaS collects reasonable reviews; the high-ticket service collects the disputes.
The debit column, documented and specific:
Aggressive and demeaning sales scripts. One complainant quotes being told that "if you can't spend EUR 3,500 you're not a normal businessman." Another documents roughly two months of continued calls after declining. One client reports verbal abuse by phone after cutting budget.
Refund refusals: EUR 1,500 kept after a cancellation within 24 hours; a EUR 2,000 partial refund refused on a EUR 5,800 program; a collection attempt where the complainant states no work was delivered.
Prices withheld until the call; payment requests without proper invoices.
The recurring arc: strong pre-sale contact, communication collapse after payment, weak deliverables.
The credit column, because an audit that omits it is a hit piece:
The company publicly responds to complaints, and several disputes are marked resolved - with a refund or a re-run campaign.
Scamadviser rates the domain "very likely safe."
Genuine positive outcomes exist in the record, including a cited 180-leads-4-sales case.
The pattern is consistent across five years and a rebrand: not an operation that takes money and vanishes, but one that sells hard at the top, under-delivers often enough to leave a paper trail, and resists giving money back until pressure is applied publicly.
The core problem: a mirror is not a source
Everything above is mechanics. Here is the actual reason this offer produces burned customers, and it would remain true even if every trust metric were honest.
The premise of the "digital double" is one equation: write a document about yourself, follow the steps, and receive an assistant that does what you do. That equation collapses at exactly one point - when the buyer's own expertise is the thing being cloned.
A digital double is a mirror, not a source. It can externalize knowledge you already hold. It cannot manufacture knowledge you don't. If you do not know what a good ad looks like, there is nothing in your head for the document to capture and nothing for the model to ground itself in - so it invents a plausible-sounding business and writes confident copy for that fiction. The digital double of a non-marketer is a non-marketer who types faster.
Now look at who this is sold to hardest: people with no marketing background. They hit what we would call double blindness. They cannot author the context, because they do not know what "good" looks like and therefore cannot write the rules. And they cannot judge the output, for the same reason - the AI's confident ad reads fine to them right up until the money is spent. Blind on the way in, blind on the way out.
For an expert, a digital double is a multiplier. For a beginner, it is a confident stranger making decisions in your name that you cannot even tell are wrong.
And this is where the funnel's economics stop being a neutral sales model. The EUR 30 workshop is structured so that its target buyer cannot succeed with it alone. When they fail, the script does not say "this was oversold." It says: you must have implemented it wrong - you need help. That manufactured inadequacy is the on-ramp to the retainer. Sell the map so the buyer gets lost; then sell the guide.
The one honest sentence that will never appear on a slide in this funnel: AI amplifies marketing skill - it does not replace it. If you do not have the skill, the AI just makes you faster at being wrong.
We have said a version of this on every AI post this blog has published: if you cannot do it yourself manually, you do not do it with AI. This case is that rule, monetized in reverse.
So what do you call it?
Not a scam, in the legal sense. A real company, real taxes, real software, a real Forbes Baltics listing, and real - if rare - satisfied customers. Calling it fraud would be both wrong and lazy.
The precise name: an aggressive-sales, disputed-delivery, high-ticket info-product and agency business, powered by a trust stack that is five-sixths inflated, teaching a technical safety claim that is dangerously false, and aimed - by structure, not by accident - at the exact audience least equipped to evaluate what it receives. The EUR 30 is not the product. The EUR 30 is the qualification fee you pay to become a lead.
If it feels like a scam to a buyer with no marketing background, that is because for that buyer, functionally, the distinction rarely matters: money out, generic deliverables in, and a phone that keeps ringing.
If you are evaluating an offer like this
Not just Sharpify - any AI-enablement offer, including ours. The checks that would have saved every complainant in this record:
Ask for the price in writing before any call. A rate card that only exists mid-call is priced to the call, not to the work.
Count the verifiable trust signals. Take the numbers on the landing page and find one - any one - you can confirm independently in five minutes. In this case the count was one of six.
Test the mirror question. Could you, given time, do the thing yourself manually? If yes, AI plus a template genuinely accelerates you. If no, no document about your business closes that gap - a human expert does, and then the tools multiply them.
Check the complaint portals under every previous company name. The registry rename history is public and free. Thirty seconds on firmas.lv would have surfaced NJ Media, and NJ Media's history.
Distrust any tool pitch that includes a magic undo. Anything that claims AI actions are consequence-free is wrong about the one thing you most need it to be right about.
The EUR 30 deck contains 27 competent prompts, one real audit framework, and a working introduction to a real tool. As a EUR 30 curiosity for someone who already knows marketing, it is defensible. As the front door to a EUR 2,800 quarter for someone who does not - the record above shows how that ends, in the complainants' own words, across five years and two company names.
The pattern survives rebrands. The registry remembers. So do we.
Disclaimer
This analysis of Sharpify was conducted by the SerpCtrl team using publicly available information - including the Latvian business registry, independent review and complaint platforms, and the vendor's own published materials - together with a practical review of the two paid artifacts SerpCtrl purchased, reflecting their technical and content state as of June 30, 2026. Supporting screenshots, documents, and source records substantiating the claims made in this article are retained. SerpCtrl has no affiliation with Sharpify, SIA Sharpify Services, or Niks Jansons, and does not sell the product analysed here (an AI "digital double" workshop or an all-in-one CRM). If you see omissions or inaccurate information in this publication, please report it in writing to info@serpctrl.lv, and we will make changes if they are justified.
/// NEWSLETTER
Get practical ideas
for digital growth
Join our newsletter for concise, useful insights on web development, SEO, and digital performance.