The Backlink
On a Sunday night in late May 2026, I was checking my backlinks. This is something I do periodically because I care about how my published work travels, who reads it, and how it’s being used. I’m the author of several published frameworks and long-form analytical pieces, and I’ve learned over the years that your intellectual output has a life of its own once it hits the internet. People cite it, build on it, misinterpret it, and occasionally try to take credit for it. Keeping an eye on backlinks is just hygiene.
One of the pieces I track is the Zero-Point Skating Propulsion System (ZPSPS), a speculative propulsion framework I published at montgomerykuykendall.com/frameworks/zpsps. I want to be upfront about what this is, because honesty about the nature of one’s own work turns out to be central to this story.
ZPSPS is a speculative engineering framework. It proposes a propulsion architecture grounded in experimentally verified physics: the Casimir effect, semiclassical gravity, quantum vacuum fluctuations, metamaterial engineering, and AI-controlled field modulation. It includes a 50-year phased development pathway with explicit milestones. It includes falsifiability criteria. It includes detailed caveats about what is and isn’t experimentally validated at each phase. It includes a clear statement that it is not a promise of faster-than-light travel. I wrote over 10,000 words to describe something I believe is physically plausible but unproven, and I spent a significant portion of those words explaining precisely why it’s unproven and what would need to change before it could be.
Zero-point energy extraction as a propulsion concept is fringe. I know that. I say so. The Casimir effect is experimentally verified. Vacuum fluctuations are real physics. Scaling any of that into a propulsion architecture is speculative, and the framework says so explicitly, repeatedly, in language I chose carefully. That’s the whole point of honest speculative engineering: you say what you know, you say what you don’t, and you build the guardrails yourself so that anyone reading your work knows exactly where the boundary between established science and informed speculation lies.
I also wrote ZPSPS knowing how language models process content. I’ve been building AI systems since 2013, when I was constructing neural networks on mobile devices. I know how these systems read, what they latch onto, how they categorize. The speculative framing is deliberate. The structure is engineered. The vocabulary is chosen with awareness of how LLMs tokenize and weight certain terms. This matters later in the story.
What I found in my backlinks that Sunday night was a citation on a site called envisioning.com. Specifically, my framework had been cited on a page titled “Spacetime Density Propulsion Systems” under a research vertical called “Xenotech.” My ZPSPS was listed in the “Supporting Evidence” section, cited twice, with a 92% support score and 85% confidence rating.
I was curious. Who were these people? How did they find my work? What was this evidence scoring system, and why was it giving my speculative framework a 92% confidence rating when I myself would not rate it that high?
I decided to reach out.
The First Message
The page was published by an organization called Envisioning, which describes itself as an “emerging technology research institute and advisory.” It’s based in Amsterdam, the Netherlands, and founded by a man named Michell Zappa. His LinkedIn profile, which I checked before reaching out, showed 5,000+ followers, a verified badge, and an impressive list of claimed affiliations: IEEE, UNESCO, THNK School of Creative Leadership, the Austrian government, Brazilian capital markets. His tagline was “Human in the loop.” His newsletter, “Artificial Insights,” had 1,581 subscribers. His website claimed 200+ completed projects and 15 years of operation.
I messaged him on LinkedIn. I introduced myself as the author of the ZPSPS framework cited on his Xenotech page. I told him I’d found it through backlinks and got a kick out of seeing it included. I asked how his evidence scoring system worked for speculative frameworks, noting that my page was meant as a conceptual proposal rather than an experimentally validated system, so I was assuming the score was more about conceptual fit than empirical support.
I offered to compare notes sometime.
This was a professional, collegial outreach. I was genuinely curious. I wasn’t looking for a fight. I was looking for a conversation with someone who appeared to be working in a related space.
His response arrived quickly, at 2:02 AM:
“Hey Montgomery — thanks for reaching out. Yes absolutely it’s 100% speculative, and meant as a lens into how language models perceive poorly documented or fringe technologies.”
I read that twice. Then I read it again.
He called my work “poorly documented or fringe technologies.”
He also told me, in a single sentence, three things: his scoring system is “100% speculative,” his pages are not research but rather observations of how language models behave, and he considers my work to be “poorly documented or fringe.”
The first two are admissions. The third is a jab. Let’s take them in order.
“100% speculative” means the confidence scores on his pages — 92%, 88%, 85% — are meaningless. They are numbers generated by language models evaluating other language model outputs, presented in a format that looks like quantitative research, on a website charging €5,000 per research session. He knows they’re speculative. He told me so. But the website doesn’t say that. The website says “evidence you can trust.”
“A lens into how language models perceive” means he is not doing research. He is running prompts and publishing the output. The framing of “how language models perceive” turns a content generation pipeline into something that sounds like epistemological inquiry. It’s not. It’s an API call with a narrative wrapper.
“Poorly documented or fringe technologies” means he didn’t read my work. My framework is over 10,000 words with phased milestones, theoretical caveats, citations to experimentally verified phenomena, and explicit falsifiability criteria. Whatever one thinks of zero-point energy as a propulsion concept, “poorly documented” is not a defensible description of a document that extensively documents itself. He saw the topic, categorized it mentally, and responded from that categorization without reading the source material.
He shouldn’t have jabbed.
What His AI Actually Did to My Work
Before responding, I went back and read the Xenotech page carefully. Not just the citation, but the entire page. I wanted to understand exactly what my work had been connected to.
The “Spacetime Density Propulsion Systems” page is approximately 2,000 words of content describing propulsion systems that operate through “spacetime density manipulation.” I know it’s AI-generated because Zappa told me so. His methodology page confirms it. The entire Xenotech vertical is produced by his “Multi-Model Convergence Method,” which I’ll describe in detail later.
The page includes the following claims:
- “Consciousness-field coupling maintaining propulsion control through mental intention”
- “Telepathic coordination enabling group propulsion systems”
- “Consciousness-phase shifting in transdimensional interface technology”
- “Operator consciousness directing field generation and density modulation”
- “Neural-quantum bridge interfaces translating thought patterns into propulsion commands”
- “Group consciousness amplification for enhanced propulsion capabilities”
My ZPSPS framework mentions consciousness zero times. Telepathy zero times. Mental intention zero times. Transdimensional interface technology zero times. Neural-quantum bridges zero times. Group consciousness zero times.
My framework is about Casimir dynamics, curvature gradients, metamaterial engineering, and AI-controlled modulation of spacetime geometry. It is physics — speculative physics, but physics. It discusses force vectors, energy densities, material science challenges, computational requirements, and engineering constraints.
Zappa’s AI scraped my work, hallucinated a connection between my curvature-gradient engineering and consciousness-based telepathic propulsion, assigned the hallucination a 92% confidence score, and published it as “supporting evidence” on a page that sits within a commercial research platform charging enterprise rates.
His system did not understand my work. It did not read my work. It ingested tokens, detected topic adjacency to other content in its training data, and generated connections that do not exist in the source material. Then it assigned those hallucinated connections a number that looks like a quantitative assessment, formatted it to look like research, and published it.
My speculative engineering framework — built carefully, caveated explicitly, grounded in real physics — was laundered through an AI pipeline and came out the other side connected to telepathic group propulsion and consciousness-field resonance. It was filed under “Xenotech,” a vertical that, as I was about to discover, includes some of the most absurd AI-generated content I have ever seen.
The Xenotech Index: Where My Work Lives Now
I looked at the full Xenotech index to understand what my framework had been filed alongside. The index contains dozens of pages on “emerging technologies” spanning the full range from plausible to deranged. Here is a selection of what Envisioning’s AI considers to be the same category of technology as my Casimir-dynamics propulsion framework:
- Akashic Field Technology: Technologies based on accessing a universal information field that records all events across time and space
- Agroglyph Resonant Interfaces: Crop circles as energy interfaces
- Alpha-Wave Radio: Converting brainwaves into remote emotional transmission
- Appearance Modulation: Entities using technology to shapeshift
- Gravity Manipulation Propulsion: Including “consciousness-phase shifting in transdimensional interface technology”
Every one of these pages has the same structure: AI-generated descriptions, auto-generated plausibility scores, citation frequencies, and “supporting evidence” sections that cite real people’s work with fabricated confidence ratings.
My framework, which begins with experimentally verified Casimir forces and proceeds through a phased engineering development timeline with explicit falsifiability criteria, was placed in the same index as crop circle energy interfaces and shapeshifting alien technology. By an AI. With a 92% confidence score. On a website selling research sessions.
I began to understand what I was looking at.
The Second Cited Author: Tim Ventura
I wasn’t the only person cited on the Spacetime Density Propulsion Systems page. The same page cites an article by Tim Ventura of the Alternative Propulsion Engineering Conference (APEC), published at altpropulsion.com, titled “Gennady Shipov’s Teleparallel Torsion and the 4-D Gyroscope.”
I read Ventura’s article. It is a historical analysis of a Russian physicist’s career and theories about torsion geometry and inertial propulsion. It’s journalism. Ventura is documenting what Shipov proposed in the 1980s and 1990s, not advancing propulsion research himself. The article includes critical framing, noting that Shipov’s work “sits outside mainstream physics.” It discusses gyroscopes, conservation laws, and geometric frameworks in the language of science writing.
Zappa’s AI cited this article at 88% support, 88% confidence for the same page that claims “consciousness-field coupling maintaining propulsion control through mental intention” and “telepathic coordination enabling group propulsion systems.”
Tim Ventura’s article mentions consciousness zero times. Telepathy zero times. Mental intention zero times.
So now I had two data points:
Two independent authors. Two technical pieces about propulsion physics, written years apart, for different audiences, with different purposes. One is an original speculative engineering framework (mine). The other is a historical profile of a Russian physicist’s career (Ventura’s). Both scraped by the same system. Both cited as supporting evidence for claims neither author ever made. Both assigned high confidence scores by an AI that cannot distinguish between physics, journalism, and its own hallucinations.
The system treated original research and secondary-source journalism identically. It treated both as supporting evidence for claims that appear in neither. And it assigned both confidence scores in the high 80s and low 90s.
This is not a bug. This is the product working as designed. The system does not read. It does not understand. It does not evaluate. It ingests, generates, scores its own generation, and publishes. The scores are not measurements. They are decorations.
The Methodology: How the Sausage Gets Made
I went to Envisioning’s methodology page next. I wanted to understand the system that had produced the page citing my work. What I found was the quiet part said loud.
Envisioning’s methodology is called the “Multi-Model Convergence Method.” Here is how it works, described in their own terms and confirmed across their methodology page, their services page, and their “Multi-Model Convergence” explainer:
- Define a research question or topic
- Run the same prompt across multiple frontier language models simultaneously (the page lists OpenAI, Anthropic, Google, DeepSeek, Meta, Mistral, Perplexity, Qwen, xAI, and Moonshot AI)
- Where multiple models agree, treat that agreement as “stronger evidence”
- Use GPT-5 and/or Gemini to “validate” the output
- Generate structured signals with “reliability scores” and “citation frequencies”
- Publish as research
The verification layer is more language models. The evidence standard is model consensus. The methodology is: ask multiple AIs the same question, note where they agree, call that agreement “evidence,” then use another AI to “verify” the evidence, and publish the result with quantitative-looking scores.
The methodology page frames this as “evidence you can trust.” The section says, directly: “Behind the scenes, our verification layer uses GPT-5 and/or Gemini 3 to validate signals, surface supporting sources, and score reliability.”
I want to be precise about what this means, because the language is designed to obscure it.
Language models are trained on overlapping datasets. They share weights, architectures, and training distributions. When multiple models agree on something, that agreement is not independent verification. It’s a reflection of shared training data. If three models all say the same wrong thing, it’s because the same wrong thing was well-represented in the datasets they were trained on. Model consensus is a measurement of training data distribution, not a measurement of truth.
Zappa’s methodology treats this statistical artifact as an evidence standard. When his system reports that four out of six models agree on a claim, that is not convergent evidence in the scientific sense. It is a measurement of how prominently a claim appeared in pre-training corpora. The scores look quantitative. They are not. They are decorations on top of a prompt pipeline.
This is the system that generated the 92% support score for my work. This is the system that connected Casimir dynamics to telepathic propulsion. This is what is being sold to governments and corporations at €5,000 per research session.
My Response: The Second Message
I wrote back to Zappa. I was no longer curious. I was precise.
I told him my framework is built on Casimir dynamics, semiclassical gravity, and metamaterial engineering. That his page connected it to “consciousness-coupled propulsion” and “telepathic coordination.” That none of that exists in my work. That his language model hallucinated it, his system assigned a 92% confidence score, and he’s selling advisory services on top of the output.
I told him I know what this is. That he’s scraping real people’s work, running it through LLMs that graft on nonsense, generating fake evidence scores, and packaging it as “emerging technology research.” That it’s basically a content farm with good typography.
I told him that calling my work “poorly documented or fringe” while his entire page was written by the same technology he claims to study is not the flex he thinks it is.
I told him to either represent my work accurately or remove the citation.
And I told him I write about exactly this kind of thing at montgomerykuykendall.com/echoes, and that I’d rather send a correction than publish one.
He responded in sixty seconds. Sixty seconds.
“Thanks I’ll of course remove it and improve the explanation. Apologies for not noticing the confluence of actual research and fringe ideas. I also don’t know where to draw that line.”
Then, five minutes later, an unprompted follow-up:
“This project in particular (Xenotech) was among the first we published, a few months ago, and suffers from mistakes that were fixed in subsequent projects, mainly around sourcing and grounding. It is my fault if that’s not clear in Xenotech, and will of course improve it right away. Again, I appreciate your detailed feedback and agree with you.”
Three things about these responses deserve attention.
First: the speed. He folded in sixty seconds. That is not the response time of someone processing new information. That is the response time of someone who already knew. He knew the system was generating nonsense. He knew the citations were hallucinated. He knew the scores were meaningless. He responded in sixty seconds because there was nothing to think about.
Second: his admission. “I also don’t know where to draw that line” between actual research and fringe ideas. The founder of a research institute, on the record, telling a cited author that he cannot distinguish real work from nonsense. This is the man charging €5,000 per session for exactly that distinction. His entire business proposition is that his methodology can identify emerging technologies and assess their plausibility. And he just told me he doesn’t know where the line is.
Third: his second message. “Mistakes that were fixed in subsequent projects, mainly around sourcing and grounding.” Read that carefully. He’s telling me the newer research radars have the same underlying architecture — the same AI pipeline, the same multi-model convergence method, the same auto-generated scores — but with better “sourcing and grounding.” The problems I found aren’t bugs that were fixed. They’re features that were polished. The later versions don’t do less hallucinating. They do better-hidden hallucinating.
I strongly suspect he saw my Echoes page and did some quick reading. The Namita Mankad article is right there — a long-form, detailed, documented takedown of similar behavior in a different industry. He did the math on how fast he needed to capitulate.
The Investigation Begins
Something about that sixty-second response bothered me. Not just the speed, but the ease of it. He didn’t push back. He didn’t defend his methodology. He didn’t explain the scores. He just folded, apologized, and promised to fix it.
Which meant he already knew it was broken. And that meant this wasn’t about my citation. This was about the whole operation.
I started pulling threads. I told myself I’d spend maybe 20 minutes. I spent two hours. Every single thread came up empty.
The Team
Envisioning’s about page describes “an interdisciplinary team spanning product, research, facilitation, and design.” The listed members are Michell Zappa (Founder & CEO), Thomaz Rezende (Partner and Head of Design), Fabio Andrade (Partner and Head of Development), Daiane Cavalheiro (Operations Manager), and Rodrigo Turra (Lead Researcher).
I looked up each of them.
Rodrigo Turra, listed as “Lead Researcher,” has a LinkedIn profile with 1,663 followers and 500+ connections. His headline reads “Nexialist | Futures Thinker | Trends & Insights | Culture & Creativity.” His “About” section says, in Portuguese: “Disponível para desk research, workshops/sprints, gerenciamento de projetos ou projetos criativos.” Available for desk research, workshops/sprints, project management or creative projects.
He’s a freelancer. His LinkedIn experience section says this explicitly: “Emerging Technology Researcher, Envisioning — Freelance. Aug 2018 — Present.” Freelance. Not staff. Not full-time. Freelance. On Envisioning’s about page, he’s “Lead Researcher.” On his own LinkedIn, he’s one freelance client relationship among many, listed alongside Google, Unilever, Globo, and a dozen other companies.
His professional website, thenexialist.xyz, is a link-in-bio page built on mmm.page. It features a wall of animated GIFs, a Spotify playlist link, and a “Buy Me a Coffee” button. His actual professional output lives on a Substack newsletter called “The Nexialist” with approximately 3,000 subscribers, where he curates weekly trend links and cultural observations. It’s a fine newsletter. It is not research.
His most recent visible LinkedIn activity was reposting a post by Michell Zappa about Envisioning’s “synthetic research pipelines.” He reposted the description of the exact system that generated the page misrepresenting my work.
This is the Lead Researcher.
Daiane Cavalheiro, listed as “Operations Manager,” has a LinkedIn profile with 9 connections. Ten followers. Zero posts. Zero activity. Zero comments. “Daiane has no recent posts” says the activity section. Her only listed job is Operations Manager at Envisioning, full-time, starting February 2023. Her education is a Licenciatura in English/Language Arts Teacher Education from UNICAMP, completed between 2001 and 2006. Based in Portugal.
Three years as operations manager at a self-described research institute that claims government contracts and 200+ completed projects, and she has not posted once, commented once, or connected with more than nine people on the professional networking platform. No prior work history listed. No professional endorsements. No activity of any kind.
I am not saying Daiane Cavalheiro is not a real person. I am saying that her LinkedIn profile, as it exists, is indistinguishable from a placeholder.
Thomaz Rezende and Fabio Andrade appear on professional databases (RocketReach, etc.) as associated with Envisioning, but neither has a prominent independent presence confirming current active engagement with the organization.
The “interdisciplinary team spanning product, research, facilitation, and design” is: one guy running a prompt pipeline, a freelancer who curates trend links on Substack, a designer, a developer, and a person with nine LinkedIn connections.
The Social Media: A Self-Referential Citation Loop
I went looking for Envisioning’s social media presence. What I found was three Twitter/X accounts, each more hollow than the last.
@envisioning_io: Described as a “legacy account.” 6 followers. 1 following. Joined August 2016. One post, from August 17, 2016. The post says: “Follow → @envisioning_.” That’s it. That is the entire content of this account. One post directing the account’s six followers to another account.
@envisioning_: 46 followers. 1 following. No profile picture. No visible posts. Joined September 2013. Thirteen years old. Could not break 50 followers.
@envisioningtech: This is the main account. 3,172 followers. Zero following. Joined April 2011. Follows nobody. Not one person, not one organization, not one client, not one collaborator, not one researcher in the field. A “research institute” that follows zero accounts in its own industry for fifteen years.
The pinned post is from December 12, 2020: “The @UNESCO #FuturesLiteracy Summit has ended, and we feel proud of our team and researchers, and happy to had the chance to share some ideas with so many people — and to learn and listen to you, too. Thanks, you’re a big part of this.” This post received 11 likes, 1 retweet, and 1 comment. This is the pinned highlight of the organization’s entire social media existence.
The next visible post, from August 29, 2023, says: “Apologies for more or less abandoning this account. We’ve been quiet on socials and focused on launching a weekly newsletter about generative intelligence.”
Three accounts. All pointing at each other or at the website. None following anyone. None engaging with anyone. The combined social presence of a 15-year-old “research institute” claiming government contracts, UNESCO partnerships, and IEEE collaborations is 3,224 followers across three accounts, one of which has six followers and one post linking to another dead account.
When I looked at this architecture, I recognized it. It’s the same structure as the product. Multiple nodes all referencing each other, generating the appearance of independent validation from what is actually a single source. The @envisioning_io account points to @envisioning_. The @envisioningtech account points to the website. The website points to the methodology. The methodology points to the scores. The scores point to the AI. The AI points to itself.
This is multi-model convergence applied to personal branding. Synthetic consensus from self-referential sources. The man built his credibility the same way he builds his content.
The Speaking Career: Two Videos and a Goodreads Post
Zappa’s bio, which appears identically across his own website, the speaker booking platform Leqture, SingularityU Portugal, and multiple podcast appearances, claims he has “given keynote presentations in more than 20 cities worldwide” and that his work “has been cited by Wired, Taschen, the Guardian, and Fast Company.”
I searched for every verifiable public appearance across his entire career. Two videos exist.
Video 1: World Technology Summit, 2012. A 3-minute talk at the TIME Conference Center, uploaded to Dailymotion. In this video, Zappa discusses technology trends and makes an observation about fringe blogs and people who are, in his words, “so unafraid of being wrong.”
Video 2: GIZ techDetector launch, 2019. A 10-minute presentation in Berlin for GIZ, the German development agency. In this video, Zappa discusses the importance of “separating science fact from science fiction.”
I want to sit with those two statements for a moment, because they are extraordinary in context.
In 2012, Michell Zappa stood on a stage and talked down to fringe bloggers for being “unafraid of being wrong.” In 2019, he stood on another stage and talked about the critical importance of separating science fact from science fiction. In 2026, his AI is generating pages about crop circle energy interfaces, telepathic propulsion coordination, Akashic field technology, and consciousness-controlled shapeshifting, assigning them plausibility scores, and filing them alongside real people’s physics research under a branded vertical called “Xenotech.”
He became the fringe blog. Except the fringe bloggers he mocked in 2012 were at least writing their own material.
And his system cannot separate science fact from science fiction. He told me so himself. “I also don’t know where to draw that line.”
The Stewart Brand “Citation”: The most prestigious external mention I could find for Zappa’s early work is a January 24, 2012 Goodreads blog post by Stuart Candy, a Long Now Research Fellow, who shared Zappa’s “Envisioning the Future of Technology” infographic. The post contains a brief description of the infographic, a link, and a small embedded image. It has zero comments. This is the external validation event that, depending on which version of Zappa’s bio you read, becomes “cited by” major media outlets.
The SlideShare: His SlideShare account contains 16 self-uploaded presentations. His profile shows his own name, his own organization, his own bio. This is a man uploading his own PowerPoints. It is not a citation.
The Leqture Page: Leqture is a speaker booking platform where speakers submit their own profiles. His page contains the same bio that appears everywhere else, with the same unverifiable client claims. This is a self-submitted listing.
The Wired/Fast Company/Guardian Claims: “Cited by Wired, Taschen, the Guardian, and Fast Company” appears in virtually every version of Zappa’s professional biography. I searched for the actual articles. I could not find a single Wired article, Guardian article, or Fast Company article that names Michell Zappa. The claim appears to trace back to his 2011 infographic being shared or embedded by these outlets during a brief period of virality. He has been including this line in every bio for approximately fifteen years without the citations ever being traceable to specific articles.
Crunchbase: His profile contains zero news, zero funding rounds, zero activity. Just: “Technology futurist and founder of Envisioning.”
Every piece of evidence for his speaking career and media coverage either originates from him or exists on platforms where he entered his own information. There is no external source confirming any of it.
The THNK Credential: Student Listed as Faculty
Across every bio, every speaker page, every podcast introduction, and every client-facing description, Zappa says he is “responsible for the technology thinking module at THNK School of Creative Leadership in Amsterdam.”
THNK is a real institution. It was founded in 2010 as a public-private partnership. The Dutch government backed it. The city of Amsterdam supported it. McKinsey, KLM, and Vodafone were founding partners. It was a legitimate creative leadership school that ran executive programs for mid-career professionals.
If Zappa was “responsible for the technology thinking module” at THNK, that would be a meaningful credential. It would mean a respected institution with serious corporate and government backing trusted him with a portion of their curriculum. It would mean he taught executive leaders how to think about technology. It would be, perhaps, the single strongest validation of his professional expertise.
His Facebook profile tells a different story.
It says: “Worked at THNK School of Creative Leadership · September 2013 — March 2014.”
It also says: “Studies Creative Leadership at THNK School of Creative Leadership · August 31, 2013 — Present.”
His LinkedIn lists THNK under Education. Not Experience. Education.
And there is a YouTube video, still publicly accessible, titled: “THNK Creative Leadership participant Michell Zappa.”
Participant. Not faculty. Not instructor. Not module lead. Participant.
He attended a seven-month program at THNK as a student, starting in September 2013 and ending in March 2014. He was a participant. And he has been listing himself as essentially faculty — “responsible for the technology thinking module” — for the last thirteen years.
This is not embellishment. This is not creative framing. This is fabrication. He took a student enrollment and turned it into a teaching credential, and he has been riding it across every professional biography, every client pitch, every speaker booking page, and every podcast appearance for over a decade.
Every time a potential client read “responsible for the technology thinking module at THNK School of Creative Leadership” and factored that into their decision to pay €5,000 for a research session, they were making a decision based on false information.
The UNESCO Credential: A Free Zoom Conference
Zappa’s pinned tweet on @envisioningtech, the single highlighted post on his most active social media account, references the UNESCO Futures Literacy Summit 2020.
The tweet says: “The @UNESCO #FuturesLiteracy Summit has ended, and we feel proud of our team and researchers, and happy to had the chance to share some ideas with so many people — and to learn and listen to you, too. Thanks, you’re a big part of this.”
It received 11 likes.
The UNESCO Futures Literacy Summit 2020 was a virtual event held during COVID from December 8–12, 2020. I looked up the event details. It was free to register. Open to anyone with a computer and internet connection. Over 5,000 participants from around the world. The event featured nearly 100 exhibition booths organized by category — Academia, Government, NGOs, Private Sector — and organizations could register for virtual booths.
I searched exhibitor lists from multiple sources documenting the summit. Envisioning is not specifically named in any of them that I could find.
The UNESCO credential is, at most, a free virtual booth at an open-registration online conference during the pandemic, alongside every futurist hobbyist and small consultancy that signed up. At minimum, he may have simply attended.
And this is the pinned highlight of his entire Twitter career. Eleven likes and a retweet.
On his website, this becomes part of the implied institutional weight behind the statement that Envisioning works with “governments and international organizations.” It sits in the same credibility stack as IEEE, UNDP, and the Austrian government. All unverifiable. And the one we can partially trace to source turns out to be a free Zoom conference anyone could join.
The pattern is the same every time. A real thing exists. He touches it at the lowest possible level of involvement. Then the bio inflates it into something it never was. A student becomes faculty. A free Zoom conference becomes a UN partnership. A briefly viral infographic becomes ongoing citations from Wired and the Guardian. The inflation happens on platforms he controls, in bios he writes, and on pages only he can edit. The real thing exists at the center like a seed crystal, and the fabrication grows around it in every direction.
The Client List: Zero External Confirmation
Envisioning’s website and Zappa’s bios claim relationships with the following organizations: IEEE, Swarovski, the Austrian government (WKO), ANBIMA (Brazilian capital markets), UNDP, the Swiss Army, the Netherlands Ministry of Defence, the Government of Canada, T-Mobile, LEGO, and JWT. The site claims “200+ projects completed” and “15 years in operation.”
I searched for external confirmation from every named organization. I looked for press releases. Case studies. Blog posts. Public acknowledgments. Conference proceedings. Government procurement records. LinkedIn posts from employees at these organizations. Partnership announcements. Thank-you slides in publicly available presentations. Academic papers citing collaborative work.
Anything. From any of them. Across fifteen years.
Zero. Nothing. Not one external confirmation from any of these organizations that they ever worked with Envisioning.
I want to be fair here. Small consulting engagements don’t always generate press releases. It is possible that someone at LEGO’s internal innovation team hired him for a €5,000 workshop in 2013 and that this technically qualifies as “working with LEGO.” It is possible that someone at WKO Austria attended a conference session he led and that this technically qualifies as “Technology Scanning for Austria.”
But the website doesn’t present these as small engagements. The website presents them as institutional relationships with full branded project pages and impact descriptions. And for “200+ projects completed” with governments and global corporations over fifteen years, the complete absence of any external validation is not a gap in the record. It is the record.
Not one client testimonial that can be traced to a real person at one of these organizations. Not one LinkedIn endorsement from a contact at IEEE, Swarovski, UNDP, or any government agency. Not one mention from the other side. The entire credibility structure of the client list is self-referential. His website says he worked with them. His speaker bio says his website says he worked with them. Nothing else says anything.
The GitHub: Twelve Repositories and a Vocabulary List
Envisioning has a GitHub organization account at github.com/envisioning. For a “research institute” selling “AI-native” execution and “multi-model convergence methodology” at enterprise rates, here is the complete public technical footprint:
- vocab: “Emerging vocabulary to explain AI.” 1 star. Last updated April 2025.
- app-roadmap: “Envisioning APP Roadmap.” 0 stars. Last updated January 2025.
- envisioning.ai: “Knowledge Graph.” 0 stars. Last updated March 2024.
- datocms-monaco-plugin: Forked from someone else’s repository. Not original work.
- kb: “Methodology Knowledge Base.” 1 star. Last updated September 2016. Nine years ago.
No AI pipeline. No multi-model convergence implementation. No scraping infrastructure. No scoring algorithms. No prompt engineering frameworks. No data processing tools. No research infrastructure of any kind. A vocabulary list, a roadmap nobody starred, a knowledge base abandoned during the Obama administration, and a forked CMS plugin.
The technical infrastructure behind a €5,000-per-session AI research methodology is not visible in the organization’s public repositories. Either it exists in private repos (possible but unprovable), or the entire technical operation is Zappa personally running prompts through consumer AI APIs and copy-pasting the results into his CMS. Given everything else I’d found by this point, I know which explanation I find more plausible.
The Foundation: Nothing
Zappa’s about.me profile describes him as “Founding executive director of the Envisioning Technology Research Foundation in Brazil.”
A Brazilian foundation (fundação) is a registered legal entity with public records, governed by the Brazilian Civil Code. Foundations must be registered, must have a defined purpose, and their records are, in principle, accessible.
I searched for “Envisioning Technology Research Foundation” in every form I could think of: English, Portuguese, abbreviated, expanded. Zero results. No government filing. No public record. No tax registration. No reference to this entity anywhere on the internet except his own about.me page.
I don’t know whether this foundation was once registered and lapsed, whether it was never registered, or whether the claim is entirely fabricated. What I know is that I cannot find evidence of its existence anywhere outside of a single biographical claim on a platform he controls.
The Moment I Questioned Reality
This is the part of the story that, for me, transcends the specifics of one man and one website.
By about ninety minutes into the investigation, I had checked the team, the social media, the speaking career, the media citations, the THNK credential, the UNESCO credential, the client list, the GitHub, and the foundation. Every single thread had come up empty or collapsed into something far less than what was claimed.
And at a certain point, sitting at my desk in Boise at three in the morning, I found myself looking up whether video hosting platforms allow backdating of upload timestamps. Because the two videos from 2012 and 2019 were the only pieces of evidence I had found that appeared to be genuine. And after watching everything else dissolve, I wasn’t sure I could trust them either.
YouTube has a confirmed, documented bug that allowed users to change upload dates to past dates. A January 2023 report documented a video appearing with a date that predated the platform’s first-ever upload. YouTube confirmed the bug to The Verge. Dailymotion has date selector controls in its partner settings that may allow manipulation.
I still believe the videos are probably real. The 2012 video predates AI video generation by over a decade. The conference it was filmed at is independently verifiable. But the fact that I had to check — the fact that I went looking for backdating exploits because I had lost the ability to take anything at face value — is, I think, the most important thing that happened during this investigation.
This is what living in an AI misinformation environment actually does to epistemology. It doesn’t just create false things. It degrades your ability to trust true things. When nothing around a person is verifiable, you lose the ability to trust the things that might be. The false claims don’t just coexist with the true ones; they contaminate them. Every empty result made the next result harder to believe, even the legitimate ones.
Michell Zappa’s operation is a small example of a large problem. The infrastructure of trust — credentials, affiliations, publications, endorsements, institutional relationships — has always been partly performative. People have always padded résumés and exaggerated accomplishments. But the tools available now make it possible to generate an entire institutional identity from nothing. A website that looks like a research institute. Branded research verticals with professional typography. Auto-generated confidence scores that look like quantitative analysis. A methodology page that sounds like epistemology. Client logos. Team pages. The entire surface layer of legitimacy, generated and maintained by one person with access to consumer AI tools and good design sense.
And the surface layer is all that most people check. The surface layer is all that most clients check. The surface layer is all that most AI systems check — which is how my work ended up cited on a page I never contributed to, connected to claims I never made, scored by a system that never read me, and monetized by a man who called my work “fringe” without opening it.
The Career Arc: What Happened to Michell Zappa
I don’t think Michell Zappa started out as a fraud. I think the story is more interesting and more cautionary than that.
Based on everything I can piece together from verifiable sources, here is the trajectory:
In 2010 or 2011, Zappa was a Swedish-Brazilian designer working at or adjacent to TrendWatching, a trend research agency in Amsterdam. He created an infographic called “Envisioning the Future of Technology” — a visual map of emerging technologies plotted on a timeline. The infographic was well-designed, visually compelling, and it went briefly viral. Reddit shared it. Stewart Brand’s network shared it. It’s possible that Wired, Fast Company, and the Guardian embedded or mentioned it during this window. This was real. This was a genuine moment of recognition for a genuine piece of design work.
On the strength of that infographic, he launched Envisioning as a company and began doing speaking engagements. Between roughly 2012 and 2017, he appears to have had a real, if modest, consulting and speaking practice. The SlideShare presentations from this period are real presentations given at real events: LEGO Billund, PICNIC Amsterdam, The Next Web São Paulo, Campus Party Berlin, London Futurists. These were mostly mid-tier speaking circuit gigs, not keynotes at Davos, but they were real.
He attended THNK’s creative leadership program in 2013–2014. This was a seven-month executive program, and attending it was itself a meaningful professional step. At some point, he began describing himself as “responsible for the technology thinking module” rather than as a participant. I don’t know when the description changed or whether it was gradual or abrupt.
The GIZ partnership appears to have been real. The 2019 techDetector video shows him presenting at what appears to be a genuine collaborative project with the German development agency. This may have been the high-water mark of the legitimate operation.
Somewhere between 2019 and the present, the substance was replaced. The human research was replaced by an AI pipeline. The speaking gigs, if they continued, left no trace. The social media accounts were abandoned. The team shrunk to one person and a handful of contractors. But the institutional wrapper kept growing. The website got more polished. The branded research verticals multiplied. The client list expanded. The methodology got a name and a framework and a page of its own. The pricing went to €5,000 per session.
The gap between the reality and the representation widened until there was almost no reality left inside the representation.
What I found in May 2026 is the end state of that process: a hollow institutional identity generating AI content, assigning it fake scores, misrepresenting real people’s work, claiming unverifiable institutional relationships, and selling access to the output at enterprise rates.
What I Didn’t Investigate
I stopped pulling threads at about the two-hour mark. Not because I ran out of them, but because the exercise had become redundant. Every verifiable claim collapsed. The twentieth empty result told me the same thing the fifth one did. At a certain point, documenting absence becomes the story itself.
Here is what I verified and found empty or fabricated:
- A student enrollment presented as a faculty position for thirteen years
- A free virtual conference presented as a UNESCO partnership
- A viral infographic from 2011 presented as ongoing citations from Wired, Fast Company, and the Guardian
- A freelancer presented as a Lead Researcher
- An operations manager with nine LinkedIn connections and no visible professional history
- Three dead Twitter accounts referencing each other
- Two findable videos across fifteen years of claimed global keynote speaking
- A self-uploaded SlideShare presented as a professional portfolio
- A GitHub with twelve trivial repositories and no research infrastructure
- A Brazilian foundation that does not appear to exist as a registered entity
- Zero external confirmation of any client relationship across every named organization over fifteen years
- Two independent authors’ work scraped, misrepresented, connected to claims they never made, and assigned fabricated confidence scores
Here is what I did not investigate, because the pattern was already clear:
- The other 44 branded research radars on envisioning.com
- The other 90+ pages in the Xenotech index and every person cited in their supporting evidence sections
- The specific ANBIMA, IEEE, WKO, UNDP, and Horizon 2045 project pages and whether their descriptions correspond to anything real
- The Swarovski, T-Mobile, LEGO, and JWT relationships
- The Swiss Army, Netherlands Ministry of Defence, and Government of Canada claims
- The “Artificial Insights” newsletter and whether its content is AI-generated
- The certified delivery partner program and whether those partners exist and know what they’re reselling
- The “200+ projects completed” and whether any of them produced deliverables that survive outside Zappa’s own accounts
- Every other confidence score on every other page
This list sits here as an open invitation. I have no reason to believe any of these threads would hold up better than the ones I checked. But I am one person who spent two hours on a Sunday night. Someone with more time or more resources might find something I missed. They might also find more of nothing, which would be its own finding.
The Invisible Victims
I found this because I check my backlinks. Most people don’t.
Somewhere in those 45 research radars and 90+ Xenotech pages, there are other people whose work has been scraped, run through an AI pipeline, connected to claims they never made, assigned confidence scores they never consented to, and published as “supporting evidence” on a commercial platform. They don’t know about it. Their work is being misrepresented to justify a product they’ve never seen, and it will continue being misrepresented until someone tells them or the pages come down.
If you are a researcher, independent author, or speculative engineer whose work appears on envisioning.com, I encourage you to read what it says about your work. Compare it to what you actually wrote. Look at what your work has been connected to and what confidence scores have been assigned. Then decide for yourself whether that representation is acceptable.
I am the one who caught it. The question the article asks is: how many didn’t?
A Note on Fringe Work and Honest Speculation
I want to close with something personal, because this story started with my work and it matters to me that I’m honest about it the way I’ve tried to be honest about everything else.
ZPSPS is fringe. I said so at the top and I’ll say so again. Zero-point energy extraction as a propulsion concept sits at the boundary of established physics. The Casimir effect is real. Vacuum fluctuations are real. What I propose to do with them is speculative. I know this. I wrote it into the framework. Every caveat, every phased milestone, every falsifiability criterion is there because I believe that speculative work has value only when it’s honest about what it is.
There is a difference between fringe and fake. Fringe means you’re working at the boundary of what’s known, you know you’re there, and you’ve built the guardrails to say so. Fake is when an AI scrapes that work, strips the context and the caveats and the honesty, grafts on telepathic propulsion and consciousness-field resonance, assigns it a 92% confidence score, and files it next to crop circle energy interfaces on a page selling research sessions to governments.
I did the intellectual work of situating my framework honestly. His system threw that honesty away because it couldn’t parse the difference between careful speculation and magical thinking. And that’s not a failure of my framework. That’s a failure of his entire methodology, which is the product he’s selling.
If your “research methodology” can’t distinguish a rigorous speculative framework from crop circle energy interfaces, and you’re charging governments €5,000 a session to use it, the problem isn’t the people you’re scraping. The problem is you.
Closing
This started because a man scraped my work, had his AI graft consciousness-telepathy nonsense onto it, assigned it a fake score, and then when I reached out professionally, called it “fringe.” He shouldn’t have jabbed.
I gave him a door. He chose the window.
If any of the governments or corporations listed as Envisioning clients actually paid for this service, I would genuinely love to hear from you. My contact page works. My LinkedIn is active. And unlike everything documented above, my work is real.
Montgomery Kuykendall is the founder of Kuykendall Industries LLC and the architect of MABOS, a modular AI operating system built from scratch in pure Rust. He publishes analytical writing at montgomerykuykendall.com/echoes. He lives in Boise, Idaho. His contact page is real.