As Seen by AI: Webinar Recording and Edited Transcript

Intentful Insights Team
October 22, 2025 at 12:00 PM
18 min read

This article presents an edited version of the webinar As Seen by AI, led by Marina Petrova, CEO and Co-Founder of Intentful. The original page includes the session recording, and the transcript has been revised lightly for readability and length. The explanations and observations in this transcript reflect Intentful's best understanding of current AI system behavior as of October 2025. While no one truly knows the exact internal mechanisms of the AI systems, the summary is based on observed behavior and hands-on testing.

Marina Petrova opened the session by framing the topic around how AI systems interpret websites and, more broadly, how they understand brands.

The webinar focused on three main areas: what matters as of October 2025, what AI systems actually see when they access websites submitted by webinar volunteers, and an agentic search demo showing how a company or destination can test how AI interprets its online presence. The session then moved into recommendations for discoverability based on Intentful’s current work and testing.

AI is not only reshaping search. It is changing the way customers, visitors, and users interact with information and content overall.

Intentful started in 2021. Since then, the company has been building AI that understands brands, destinations, and companies. Intentful works globally with organizations across tourism and travel, performing arts, CPG/FMCG, telecommunications, agencies, and other expanding sectors.

The guidance shared in the webinar comes from Intentful’s direct daily work with AI. Because Intentful’s products rely on AI search, the recommendations are grounded in hands-on product development rather than commentary gathered from general online sources or thought leadership.

Grounded in Real Expertise, Not Theory

Although much of the session addressed search, Marina emphasized that the larger shift is about how information itself is handled. The well-known changes in search behavior and traffic patterns are only part of the story.

AI Responses Draw from Hundreds of Sources

In the previous search environment, optimization was often centered on a company or brand’s core website. In the current AI environment, a single answer to one query can involve hundreds of sources, and sometimes even more. That means web presence still matters, but the brand story is now shaped across many touchpoints where the brand may appear.

Marina noted that one of Google’s founders has referenced thousands of information sources in a single query analysis. Whether that was meant literally or to indicate scale, Intentful would not be surprised if AI systems evaluate thousands of sources before returning an answer.

This changes the job beyond website optimization alone.

If a brand does not supply strong signals, AI will describe that brand based on what it can find rather than what the brand intends to communicate. That matters for content programs, PR campaigns, and every other context where the brand is mentioned.

At a simplified level, the webinar identified two urgent goals:

• Make sure AI systems can discover the brand.
• Influence and own the story AI tells about the brand.

Goal One: Discoverability

While preparing for the webinar, Intentful received website submissions from companies and destinations that volunteered for analysis. One destination and its agency later asked not to be included, because the way AI described the destination did not match how they wanted to be represented. That example illustrated how much work may be needed to take ownership of the AI-generated brand story.

Content remains extremely important, and Marina emphasized that she has said this for years. But content is not the first place to begin. The first priority is discoverability.

Most websites were built for an earlier playbook and are therefore not optimized for AI systems. Structurally, many sites are not visible to AI. Large amounts of content cannot be read by AI systems at all, as the webinar demo showed.

Three foundational points were highlighted.

First, websites must be open to AI bots and AI crawlers. Some organizations avoid allowing AI bots because hosting costs can rise and because AI crawlers behave differently from Google. Those concerns are real, but they can be managed. Site owners can set rules about how often bots are allowed to visit. The first step is ensuring that AI systems are not completely blocked, because otherwise they cannot add the site’s information to their knowledge bases.

Among the volunteer websites reviewed by Intentful, some blocked about half of the relevant bots. For several EU-based companies, the sites were fully blocked from AI systems. In those cases, no matter how strong or polished the content may be, AI cannot access it.

When Intentful refers to AI in this context, it includes bots from OpenAI, Perplexity, Anthropic, and Google. Traditional SEO continues to matter for several reasons, but discoverability now also means being findable by AI systems, not only by traditional search.

Every website now has two audiences: people and AI systems.

How AI Sees a Website Compared with How People See It

The examples in the webinar came from Destination Marketing Organizations, DMOs, which promote destinations to visitors and support local communities.

Marina described DMO websites as visually beautiful and often filled with color, imagery, style, and care. For human visitors, these sites can feel clear, organized, and inspiring.

The webinar showed screenshots from volunteer sites that looked rich and structured to people. Then it compared those views with what an AI system could read from the same pages.

One detailed example focused on Oconomowoc, Wisconsin. The site had typical destination content: things to do, places to stay, dining information, and related visitor resources. For a person, the homepage appeared polished and inviting, with embedded live video and strong visual appeal.

The AI view of that same homepage was very different. What a human visitor sees is not necessarily what an AI system can access.

Another example looked at a Things to Do page with substantial destination information, including beaches, fishing, water activities, boat rentals, theater, concerts, and sports. For an organization of that size, the content represented significant effort and value. AI could read more from that page than it can on many DMO sites, where it may only find a paragraph or a few lines. Even so, AI still saw only a portion of what was actually available to human users.

When an AI bot reaches a site and cannot find usable information in seconds or milliseconds, it moves to another source. If the needed information is invisible, the site loses the opportunity to inform a potential visitor.

A further example came from a section called Water. In that case, the AI system could detect only the page title, followed by an otherwise blank page.

Across multiple volunteer websites and destinations, the same pattern appeared frequently: beyond the header, AI found either nothing or very little.

Discoverability is therefore the necessary first step. It ensures that the time, budget, and creative energy invested in content can serve both human audiences and AI systems. Advertising can still drive traffic, but for organic visibility, pages cannot appear empty to AI.

Agentic Search Demo in ChatGPT

The webinar then moved to a demonstration of Agentic Search in ChatGPT.

Marina used a pre-recorded video made the day before the webinar to keep the example recent. The query was fictional and described a conference context, even though Marina was not actually speaking at such a conference. Because it was run inside her ChatGPT Business account, the system already had context about her.

The prompt asked for recommendations for a first visit to the destination, including where to stay and the best places to eat.

The important difference from a standard ChatGPT search was the use of agentic mode. That mode reveals more of the system’s reasoning and gives insight into how it makes decisions, what it opens, and why it moves from one source to another.

The same destination, Oconomowoc in Wisconsin, was used. Marina encouraged attendees to try the process for their own organizations or destinations. She recommended recording the session so it can be paused and reviewed in detail.

What the Agentic Demo Revealed

• The model first chose to open the official DMO website and treated it as a credible local source for lodging information beyond TripAdvisor. This was notable because newer AI systems can distinguish official sources from unofficial ones in ways earlier versions did not.
• It reviewed the dining section and noticed that the main page showed restaurant names but little additional information. The model then followed restaurant links to seek more detailed listings.
• When the site could not provide what the model needed, it returned to search results and began opening individual restaurant websites.
• It reviewed those restaurant sites and observed that even simple pages could contain useful sections such as Our Story and Local Partners, which it opened for added context.
• The full sequence ran much faster than similar searches used to run: about 5 minutes rather than 10–15, while still gathering and combining information from multiple sites.
• The model then visited Travel Wisconsin, the state tourism website, but recognized that automated access was blocked. It quickly stopped trying that source and moved to accessible sites such as Twisted Fire.
• On the Twisted Fire site, it observed that extensive image use limited the amount of readable text. This was important because it showed that strong content still depends on technical accessibility, crawlability, and readable structure.
• The model also attempted MapQuest and other sources, periodically returning to Travel Wisconsin despite the access block, showing persistence in its reasoning.
• It opened Yelp but could not extract usable content because of dynamic loading. After identifying that limitation, it left and moved to a restaurant’s own site, Badger Burger.
• On the Badger Burger site, the model encountered a long page, went to code line 888, and processed the page quickly. This showed how efficiently AI can scan and interpret extensive content.
• It evaluated language such as flavorful burger ingredients and considered whether that information would help travelers, then returned to the DMO site for additional context.
• The overall process demonstrated how agentic AI can resemble human research behavior by exploring, comparing, checking, and synthesizing sources, but with far greater speed and scale.
• The final result was a coherent summary that blended information from the DMO, restaurant, and travel-related sites.
• Running and recording this type of agentic search for a business or destination makes it possible to see what AI can read, what it skips, and which sources it prioritizes.

Marina did not ask follow-up questions in the demo because the purpose was to show how the system searched. In a real user scenario, visitors would likely continue with more specific questions about their own trip or need.

She strongly encouraged organizations to watch how AI evaluates their own websites and which other information sources it uses.

Three Steps to Help Your Website Appear in the AI Source Set

The webinar then turned to the practical question of how to make sure a site is included when AI systems gather information.

Marina outlined three core steps. She noted that some attendees may have heard this framework before, and that it remains relevant.

Step 1: Discoverability

Organizations should turn discoverability into a checklist and review it with their web team or developers. The team should be able to show whether the site is open to Perplexity, ChatGPT, Google, Apple, AI, and other relevant systems; whether anything is restricted; and if restrictions exist, whether bots are still allowed to access the site on a controlled schedule such as once a day or three times a day.

For sites with significant traffic and concerns about hosting costs, current robot.txt rules should be reviewed. A development team can explain what is allowed. A non-developer can still understand enough to discuss whether the rules make sense.

Sitemaps require particular attention. Marina said that in 95% of cases, sitemaps are disorganized.

The first requirement is having a sitemap at all. Across the sites Intentful has reviewed, approximately 35% do not have a sitemap.

Once the sitemap is accessible online, teams should spend time reviewing which pages appear in it. A sitemap is not only a site map for people and search engines; it is also a map that helps AI understand which content exists. These reviews often reveal large amounts of outdated or irrelevant content. In the older search environment, keeping some old pages could make sense because someone might land there and continue to another page. That logic no longer applies in the same way. Content should be fresh and updated.

Teams should also confirm that all important pages are included. Intentful often sees pages that have been published but are missing from the sitemap. Unless ads drive traffic directly to those pages, they may never be found. Paid traffic can be useful, but organic discovery should not be ignored.

After reviewing one sitemap, the process becomes easier to understand. Teams should clean it up, keep it current, and understand whether it is updated manually, automatically, or through another process.

Accessibility also matters. In this context, accessibility refers to whether the website follows accessibility standards as well as possible, because this affects what AI can interpret.

The ChatGPT agentic reasoning demo also showed the issue of dynamic rendering. Marina was not suggesting that all dynamic loading elements should be removed. Instead, teams should discuss whether dynamic elements can also be exposed in a static way so AI can recognize that the content exists. If an agent is looking for tickets, hotels, or dates, it may click into dynamic content, but it first needs evidence that the content is present rather than seeing a blank page.

Step 2: Structure and Signals

The second step is structure and signals. Much of this comes from traditional search and basic SEO hygiene.

If a page loads too slowly, AI may skip it and choose another source. PageSpeed Insights and related Google tools make it possible to check speed and performance without being a technical specialist. If the site falls outside guidelines, the next conversation with developers should focus on improving load time. This connects to the demo example where the model avoided a page because images were too heavy.

Basic SEO tags still matter. Open graph tags still matter. Structured data is especially important because it helps AI read and understand content. This is not only about ChatGPT or Perplexity; structured data has long been important for Google and will continue to be important.

Structured data should be treated as seriously as the sitemap. Based on the organizations registered for the webinar, many had events or other content types that require structured data. Google provides documentation on available structured data types, and developers or web teams can add it relatively easily.

Semantic markup is also important. Sites should avoid unnecessary pagination when it hides content behind additional clicks.

Step 3: Content Interpretation

The final step is content. Content remains extremely important.

Keywords no longer function in exactly the same way they did in traditional search, but they still provide signals. Teams should not ignore them completely.

Text content is essential. AI can read images, view images, and watch videos, but text remains the primary entry point. Pages should include descriptive, clear information about the subject they need to cover. The language should not be only promotional; it should also provide useful detail.

Freshness is another major factor. AI systems increasingly check for the most current information they can access. Intentful still sees cases where content points back to 2016.

The context window will also continue growing. In earlier systems, AI considered shorter amounts of text to understand a topic. Now it can work with a much larger volume, comparable in simplified terms to a book-length amount of text.

This creates two content streams.

The first is inspirational content for humans. People still want visuals, videos, color, emotion, and inspiration.

The second is informational content for both people and machines.

That informational content needs to be specific. If a detail is known internally but not published on the website, AI will not know it. AI also makes it possible to move beyond assumptions about user needs and instead understand real intent.

Marina described the current moment as an opportunity to connect with every customer, user, and visitor in a way that was not possible before.

In a prior webinar held about a couple of months earlier, Intentful reviewed a sample of 15,000 questions from destination websites. Those were questions visitors asked through the Intentful AI Assistant installed on destination sites. Marina did not repeat that full analysis, but she referenced it as a powerful way to see what people are truly asking about.

There is no PII attached, we do not collect personal information, so this is all anonymized. Even without knowing who asked each question, the dataset provides rich insight into what visitors seek when they arrive on a website.

In the DMO context, teams should ask whether their websites contain enough information for AI to answer visitor questions, whether through an on-site AI assistant, a ChatGPT AI bot, or Google.

People use AI as though they are consulting a trusted local, not interacting with a technical system. They ask about hours, parking, average costs, restaurants, and many other practical details. They need information, not only marketing inspiration. Content strategy should therefore become more detailed. Marina acknowledged that this requires significant work, but if organizations want visitors to find and use the information, the content must be updated.

The broader shift is also larger than search. Companies are moving away from speaking at customers through websites and ads and toward speaking with customers. This is the beginning of two-way communication.

Summary: Knowledge and Content

After discoverability is solved and AI no longer sees the website as blank, AI works from the knowledge a brand publishes.

At the time of the webinar, AI breaks information into chunks. That will change as context windows grow, but for now, systems split content into smaller manageable units, understand the context, and reassemble the most relevant pieces when answering a question.

To be included in those answers, organizations need to make sure they are discoverable and that the relevant information is available on the website. Professional knowledge should guide content choices, but customer questions should guide them too.

A content strategy for people and machines should balance inspiration with information. People frequently ask in-the-moment questions. Intentful sees repeated examples of users asking what is happening right now, sometimes in live situations such as being stuck in traffic and assuming AI will know.

For DMOs especially, member pages need attention. A restaurant name alone is not enough, and an event listing should include more than a name. Pages should provide as much useful information as possible, not only marketing language. Marina recommended thinking like a trusted guide or front desk resource, not only like a marketer. Information should stay current, because content from several years ago is increasingly unhelpful.

The Four Buckets: A Practical Checklist

Intentful organizes the work into four buckets:

1. Discoverability. This is the first priority. Organizations can see the issue directly by using ChatGPT, turning on agentic mode, and observing what the system knows and can access.
2. Structure and signals. Sites need structured data, a sitemap, and related technical foundations. Web teams should be part of this process. For many developers, AI search is still relatively new, so this is not about assigning blame. It is about engaging them in the new requirements.
3. Content. The website should include both inspiration and practical information. Teams should review real user questions, create informative content with meaningful depth, and keep it fresh and accurate. They should also remember the shift from speaking at customers to speaking with them.
4. Ongoing brand monitoring. Organizations should keep reviewing how AI describes the brand across many sources, not only the website.

Intentful Products for Discoverability and Engagement

Intentful offers two product categories for customers.

The first focuses on discoverability: how AI interprets a brand or destination. This program is called As Seen by AI©. It was recently launched as a 12-month program, and Intentful is already working with several DMOs. Marina emphasized that this is not a one-time project, because AI is changing continuously and because the discovery process can reveal many website changes needed before AI can properly read the site.

The second category focuses on applying AI to user engagement. Intentful’s AI Suite includes an AI assistant, generative response ads that allow people to have a real-time conversation with an ad, and an on-brand content tool powered by the same AI that understands the brand.

Marina closed by encouraging organizations to begin with discoverability, because the foundations remain important even as AI continues to change. Strong SEO is already a useful starting point for AI search optimization, but there is additional work to do. The essential first move is making sure AI can see the website, so the brand can regain control of how it is represented.

About This Article

Intentful Insights covers AI, brand strategy, and the shift from speaking at customers to speaking with them.

This article is an edited transcript from the webinar As Seen by AI, covering how AI systems read and represent brands online, the gap between what humans see and what AI sees on websites, an agentic search demo in ChatGPT, and actionable recommendations for improving AI discoverability, structure, and content. Intentful's As Seen by AI© program and AI Suite products are referenced.

Intentful is commercially deployed since 2021, working with organizations in travel and tourism, performing arts, CPG, telecommunications, and agencies globally. Contact: [email protected]