Video UGC is becoming machine-readable GEO content. VideoWise launched tools that convert TikTok, Instagram, and user-generated video into structured assets AI systems can retrieve and cite, including YouTube GEO. On episode 18 of The GEO Show, recorded Friday, 4 September 2026, Paris Childress calls that unlock the day's biggest story, then walks Gemini 3.8 citation bugs, Treyci engines disagreeing 81% versus 43% on brand presence, The HOTH's 22.79% versus 2.45% AI-referred conversion gap, schema evidence that fails to lift citations, Profound and AirOps platform shifts, and Cloudflare's September 15 default AI crawler block.
Video UGC is becoming machine-readable GEO content. VideoWise launched tools that convert TikTok, Instagram, and user-generated video into structured assets AI systems can retrieve and cite, including YouTube GEO. Proprietary knowledge trapped in webinars, demos, customer interviews, and SME clips stays invisible until it is transcribed, structured, and made crawlable.
Paris Childress recorded episode 18 of The GEO Show on Friday, 4 September 2026. He called VideoWise the biggest story of the day, then walked the infrastructure layer around it: Gemini citation bugs, cross-engine disagreement, a 10x AI-referred conversion gap, schema evidence that fails to lift citations, Profound and AirOps platform shifts, and Cloudflare September 15 default AI crawler block.
Because most brand knowledge still lives in formats AI cannot cleanly retrieve. VideoWise is designed to turn TikTok, Instagram, and user-generated video into structured content that models can cite, including YouTube GEO.
It is absolutely essential to transcribe and structure all of your videos.
Paris framed the lesson for SaaS especially: webinars, customer interviews, demo videos, and SME videos have to become factual crawlable assets. Pure video is no longer a safe place to hide information gain.
Yes. Early Gemini 3.8 Flash testing inside Google AI Mode showed many top-of-funnel answers with no links or citations. Google Search VP Robbie Stein said that behavior is not working as intended and a fix is coming.
Share of voice measurements can get disrupted by model changes and bugs that have nothing to do with GEO work. Annotate those windows. On September 3, ChatGPT, Claude, and Grok also experienced outages in roughly the same morning window with no confirmed shared cause. Treat engines as infrastructure and annotate odd measurement periods the same way.
Treyci analyzed more than 1,200 scored AI answers. In one B2B software category, tracked brands appeared in 81% of answers from one engine versus 43% from another. Engines can disagree by nearly 2x on whether a brand exists.
Paris takeaway was to report engine-level share of voice and weight each engine by where buyers actually research. Botify competitor mention and citation share of voice, including domains that generate competitor citations, points the same direction: measurement is commoditizing, and value is moving to the execution layer that changes share of voice.
The HOTH 12-month field study across dozens of campaigns says its strongest citation gains came from technical accessibility, deeper commercial content, earned media, and authority. On its data, AI-referred visitors converted at 22.79% versus 2.45% for standard organic, roughly a 10x gap.
GEO looks less like a replacement for SEO and more like a different weighting of authority signals. Separately, Search Engine Land reviewed evidence including an Ahrefs test of 1,885 pages adding JSON-LD and found no meaningful citation uplift across ChatGPT, AI Mode, or AI Overviews. AI Overview citations declined 4.6% in the tested cohort. Schema remains useful for machine clarity. Current evidence does not show it drives citations.
Profound published Context Manager Structure on September 1, structuring business knowledge into brand, personal, knowledge-base, and synced-data context, then auto-distilling six brand records. AirOps is repositioning from pure software toward platform plus channel expert services tied to pipeline. Moonshot AI Kimi ecosystem is preparing for a possible Hong Kong IPO that could raise about $3 billion. Market-specific engine coverage matters, especially in APAC.
Cloudflare policy, effective September 15, will by default block training and agent category AI bots on ad-monetized pages for new domains. Multipurpose crawlers such as Googlebot, Applebot, and Bingbot can get blocked when customers choose to block training. Paris policy is to allow legitimate bots, watch for spoofing, and manage the new default consciously. Crawler access can change AI visibility independent of any content or SEO work.
Opening. Gemini 3.8 Flash missing citations in AI Mode. Robbie Stein: not working as intended. Annotate share of voice.
Treyci. 1,200+ scored answers. 81% versus 43% brand appearance across engines. Single-engine scores are almost meaningless.
The HOTH. Fundamentals drove citation gains. AI-referred conversion 22.79% versus 2.45% organic.
VideoWise. Paris biggest story: unlock video knowledge into structured, crawlable GEO assets.
Schema evidence. Ahrefs 1,885-page JSON-LD test: no meaningful citation uplift; AI Overview citations -4.6%.
Cloudflare. September 15 default blocks training/agent bots. Multipurpose crawler caveat can hit Googlebot, Applebot, and Bingbot.
The throughline stays the unlock and the infrastructure around it. Structured knowledge and crawler access move AI visibility as much as another dashboard.
Hi, everybody, and welcome back to the GEO Show, brought to you by GEOforge, full self-driving for AI visibility. I'm your host, Paris Childress, and I'm coming to you on Friday, September fourth with episode eighteen. Let's get right into today's top stories. Google admits that Gemini 3.8 is missing citation behavior and that this is a bug.
Early testing of Gemini 3.8 Flash inside of Google AI mode showed many top-of-funnel answers with no links or citations. Google Search VP Robbie Stein responded today that this is not working as intended, and he said that a fix is coming. So always be aware that your share of voice measurements can be really disrupted by model changes and bugs that may have nothing at all to do with your GEO work.
So it's important that you keep track of this and you annotate this in your reporting. All right. Next, the next story. AI agents can disagree by nearly two X on whether a B2B brand exists. A company called Treyci, analyzed more than twelve hundred scored AI answers, and in one B2B software category, tracked brands appeared in eighty-one percent of answers from one engine versus just forty-three percent from another.
we don't know which engines they are, but, again, this is clear evidence that a single visibility score from, let's say, one engine like ChatGPT is almost meaningless. It's very important to report on engine-level share of voice and, then try to weight each engine by where the client's buyers are actually conducting their research. 'Cause we see a massive variance of visibility across different AI engines.
All right, the next story. A twelve-month GEO field study finds no special AI hack. The HOTH, H-O-T-H, says that its strongest citation gains across dozens of campaigns came from familiar fundamentals, which are technical accessibility, deeper commercial content, earned media, and authority. It reports AI-referred visitors converting at twenty-two point seven nine percent versus just two point four five percent for standard organic on its own data.
that's massive. I mean, a ten X difference in conversion rate is really massive, and they're bringing it back to the fundamentals, technical accessibility of the content, going deeper with that content, going further down the long tail, and addressing all of the questions and pain points across all of the micro personas that exist, and of course, doing really good earned media and authority building off-site.
So GEO here appears less likely a replacement for SEO and more of a different weighting of authority signals altogether. Next story, Botify adds direct competitor citation intelligence. Botify added competitor-level mention share of voice and citation share of voice, including which domains generate competitor citations.
It also in- introduced semantic grouping of related prompts and topics. So again, we are seeing that measurement alone is not enough. It's not a moat. AI visibility monitoring is becoming a commodity function inside of established SEO suites. So competitors are winning on and adding value on what happens next, which is, the execution layer, winning citations and trying to determine and analyze what actions are actually changing your share of voice The next story is probably the biggest one we have today.
Video UGC is being turned into machine-readable GEO content. Reported yesterday from, interestingly, the National Law Review, the company VideoWise has launched AI visibility tools designed to convert TikTok, Instagram, and user-generated video into structured content that AI systems can retrieve and cite.
Its offering includes YouTube GEO and AI search visibility tooling. I think this is fascinating. this is real evidence now of, a massive unlock of brand knowledge, proprietary data, proprietary knowledge, information gain that currently is inaccessible in pure video form but can now be unlocked and fed to models for training purposes.
And that's-- this is really a lesson that I think for all companies, particularly SaaS, it's absolutely essential to transcribe and structure all of your videos. So that could be, webinars, customer interviews, demo videos, SME videos, and make sure that they are all factual, factual and that they become crawlable assets.
And VideoWise seems to be a tool now that allows for that to happen. So that is a really big unlock of information gain for many, many brands. Next story, which I'm sure most of you heard about and experienced directly. Yesterday, on September third, ChatGPT, Claude, and Grok went down all- almost simultaneously.
They all experienced outages within roughly the same morning window. But there isn't really a common cause so far that has been identified. OpenAI attributed it-- attributed its outage to a routine error. Anthropic and AI reported service dis-disruptions, but no shared cause has been confirmed. So, AI engines are now infrastructure.
this, this is, this was major news and of course, like any infrastructure, there could be failures. And we also have to s- connect this with GEO measurement platforms and, and make sure that we annotate these types of odd behaviors, just like we saw with Gemini 3.8 Flash. Next story. New SEO analysis further weakens the case for schema as a GEO layer. Search Engine Land reviewed recent evidence, including an Ahrefs test of one thousand eight hundred and eighty-five pages adding JSON-LD
It found no meaningful citation uplift across ChatGPT AI mode or AI Overviews. In fact, AI Overview citations actually declined four point six percent in the tested cohort. Separately, cited pages on AI Overview SERPs generated roughly one hundred and twenty percent more clicks per impression than uncited pages.
So this is really debunking one of the most popular best practices in on-page technical GEO, which is your JSON schema.org snippets. this is widely considered to be best practice, even essential. and I do think that schema remains very useful for machine clarity, but the evidence says that it doesn't necessarily drive citations.
There is not a direct correlation with, schema presence and AI citations. So we wanna keep monitoring this over time. that said, I think it's still a good idea to include your schema.org snippets on all pages ju-just to assist, AI bots in their understanding. Next up, China's Kimi ecosystem is preparing for a potentially three billion dollar IPO.
Reuters reports that Moonshot AI, the developer of the Kimi family of models, confidentially filed for a Hong Kong IPO that could raise about three billion dollars, with the company valued at around fifty billion dollars in an ongoing funding round. So, these Chinese models are, op-- these open source Chinese models are really serious.
I think they're closing the gap with frontier models, and we can't just assume that AI... the AI search market is gonna remain a four-engine Western ecosystem. So we, we always need to be keeping an eye on these, these, these models coming out of China, particularly if you're, if you're, doing any marketing in APAC. we need to think beyond just, ChatGPT, Gemini, Claude, and Perplexity, and start to think through the lens of market-specific engine coverage. All right. Let's move along to the next story
which is about Profound. Profound published on September first on their blog, which was corroborated by Search Engine Watch on September third, something called Context Manager Structure, which structures business knowledge into four distinct areas: brand context, personal context, knowledge base, and synced data.
And it includes Gong, Slack, and email feeds, and it auto-distills all this data into six brand records. And those are company, market, offerings, audience, strategy, and execution. It proactively asks clarifying questions instead of requiring manual dashboard upkeep. So it's enterprise only at launch, and self-serve customers must request a demo.
So Profound is moving, very interestingly. Profound is the, I think the clear head and shoulders leader in the GEO category. They've raised the most money, and I think they're the, the first unicorn in the space valued at over a billion dollars. They are moving from pure measurement towards something like an AI marketing brain that is ingesting propri-proprietary business data from as many places as it can to make its agent recommendations more specific.
So, this is actually a really direct architectural analog to what GEOforge's BaseForge knowledge ingestion work is also trying to do. We are trying to identify and absorb as much proprietary data into a knowledge base as we possibly can, sync it and automate it to make it easier, and then restructure that and deliver it back to AI engines in the form of published content on owned properties and on earned properties so that we can effectively, update and refine AI's understanding of accurate information about that brand So, Profound is, is really moving into structured business knowledge, which downstream leads to better AI recommendations.
Next story, let's talk about AirOps, another big player in the space. they seem to be repositioning from a software tool to a hybrid platform plus services provider. So they're moving into the agency space. their CEO, Alex Halladay, published on September second, a major blog post entitled 'The Next Chapter for AirOps.'
And in it, he claims AirOps is combining its platform with channel expert services teams, arguing that AI search visibility now span-spans five channels: owned content, earned content, AI ads, community discussion, and social/influencer voice. And all that must tie to pipeline revenue, not just visibility scores.
I could not agree more. he cites proprietary data that they have, which includes six billion citations, forty-seven million actions, one point five billion prompts tracked. He also cites that eighty-six point six percent of marketing leaders prioritize AI search, but only twenty-three point two percent feel confident measuring its impact.
that-- this is really telling. there's a lot going on here. AI, -- I'm sorry. AirOps is sitting on a mountain of data that they're collecting in their platform, and what they're seeing is that the complexity of this work dictates managed services more and more. And I think a large part of their offering is going to be bundling services within the platform itself.
And this is something that we are also, doing very actively within GEOforge. So their-- AirOps is, is moving toward a we do the work for you model rather than a pure self-serve software model. And I think this is a bid to capture more budget from agencies and from in-ho-in-house teams that lack direct GEO experience.
And it's also a hedge against tool commoditization. I think it's a really savvy move that we also ought to really pay attention to here at GEOforge. In fact, the, the business plan that we have is exactly that. We have a platform that we offer direct as a, as a software license, and marketing teams can use it.
agency partners can also use it, and they can layer their m-managed services on top. We also have a managed services team in-house that can effectively do this on behalf of the customers. All right.
The next story, let's go to Cloudflare. This is gonna be our last story of the day, and, also I think this is a very, very important update. Cloudflare is by default going to be blocking certain AI crawlers effective September 15th. That's coming up quite soon, in, in 11 days The policy which is-- which was first announced on July first, but its effective date is September fifteenth, says that Cloudflare will by default block training and agent category AI bots, but not search category bots on ad-monetized pages for new domains onboarding to Cloudflare.
Existing customers can opt out beforehand. Notably, multipurpose crawlers such as Googlebot, Applebot, and Bingbot will get blocked for customers who choose to block training since those bots serve dual functions. this is really, really important now because I think Cloudflare is, as I've stated in previous episodes, it's an absolute must-have tool for monitoring AI bot crawl activity on your website.
This is the first, precursor to getting any AI search visibility. It's seeing how AI bots are crawling, how often they're crawling, how deep they're crawling, which pages they're prioritizing, and this is the strongest early indicator of which pages are gonna be winning AI citations and could potentially deliver AI-referred traffic.
So it's a direct and dated technical change to crawler access that could measurably affect which brands content AI engines can retrieve and cite going forward. And this is really precisely the kind of infrastructure shift that changes AI visibility independent of any content or SEO work. So, I think it's very important to understand that any brand that is using Cloudflare and who blocks training bots without understanding the multipurpose crawler caveat could inadvertent- inadvertently block Googlebot, Applebot, and Bingbot too.
So our policy in general is to allow all legitimate bots to crawl. We don't wanna block any of them because we wanna give them full access to update their training models and their training datasets with our proprietary data. w- what we do wanna spot, though, is we wanna spot the, spoof- spoofing bots that are mimic- mimicking the real bots, and they're there to try to perhaps do malicious things or, or steal some sensitive data.
So we are monitoring Cloudflare AI bot activity very, very closely, but we are not blocking by default any, training or agent bots. But it's important now that this has to be managed because the default setting will be to block those bots effective September fifteenth. All right. That'll do it for today's episode, and thank you for hanging with me, and we will see you all in the next one.
