At HumanX 2026, the conversation wasn't about whether AI would transform how businesses reach customers. That question has been settled. What dominated three days of panels, meetings, and hallway conversations in San Francisco was something more urgent: whether companies are building the infrastructure to be found, cited, and trusted in a world where AI systems — not search bars — are increasingly the first stop in the customer journey.

This matters now because AI is no longer just assisting search—it is actively filtering, answering, and recommending, often before a user ever clicks on a link.

AiNews.com CMO and Head of News Reporting Alicia Shapiro moderated 3 panels at HumanX, covering the full arc of that transformation: how AI is reshaping the customer journey in real time, what it takes to market AI tools in a skeptical market, and how retail AI is learning to understand and predict human taste. Across all three conversations, a single pattern emerged — the companies moving fastest aren't the ones with the biggest budgets or the most technical teams. They're the ones willing to rethink the system from scratch.

This change impacts every organization that relies on discovery—from retailers and SaaS companies to publishers—because visibility is no longer driven by rankings alone, but by whether AI systems can interpret and recommend what they offer.

In short, HumanX made one thing clear: the intelligence layer that sits between consumers and the internet is no longer theoretical. It is live, it is making recommendations, and it is already deciding which brands get found and which ones don't.

The central question at HumanX was not whether AI changes the customer journey — it was whether businesses have built the data infrastructure, organizational alignment, and trust signals required to show up in a world where AI answers before anyone clicks.

AI-driven customer discovery now relies on intent-based queries, structured data, real-time signals, and trust to determine which products and content get surfaced by AI systems.

For 25 years, the keyword was the foundation of how businesses understood what customers wanted. Bill Gross, founder and CEO of ProRata AI — and the entrepreneur whose company goto.com licensed paid search technology to Google before Yahoo acquired it — put the transition in stark historical terms during the first panel, Powering the Full Customer Journey.

"For 100 years in the last century, all of marketing was focusing on demographics," Gross said. "And then starting in the late 90s, the keyword became the new measure for how customers were focused. But for the last 25 years, the keyword has been the focus of everything. All of a sudden, starting with ChatGPT — people have now changed from the keyword to asking questions."

The difference isn't cosmetic. A keyword captures a moment — someone searched, something happened. A question reveals the reasoning behind the search: the context, the need, how close someone is to a decision. That's forward-looking information, not historical. Gross drew a direct analogy: looking at a click is like taking a flashlight to the endpoint of a path. Reading the full sequence of questions someone has asked is more like an MRI of the customer's mind. "You can almost read someone's mind by looking at their whole process of all the questions they've asked," he said.

Kate Prouty, CIO of Akamai — which processes a significant share of real-time global internet traffic at the edge — reinforced that the implications go far beyond the marketing department. The customer journey is no longer linear. It moves constantly between marketing, sales, support, and product. A salesperson reaching out to a customer who just opened an angry support ticket isn't just awkward — in today's environment, that kind of disconnect is visible and damaging in ways it wasn't before. It shows that the company isn't paying attention.

"Gone are the days when marketing and sales and services were sort of loosely connected with these handoffs in stages," Prouty said. "The process now is much more conversational, much less linear. These teams really need to start working together in a way that they haven't before."

One of the sharpest moments in the first panel came when the conversation turned to dashboards — a technology most enterprise companies have invested millions in building.

When asked directly whether dashboards still matter, Prouty didn't hesitate. Prouty's verdict was unambiguous: "I will go so far as to say no." Dashboards, she argued, are built on retrospective data — breadcrumbs, cookies, click patterns that tell you what a customer did, not what they're about to do. In a world where AI can process real-time behavioral signals across all customer touchpoints simultaneously, the dashboard's core function has been superseded.

The replacement isn't another tool. It's a different operating model entirely. Prouty described Akamai's own internal restructuring — building new organizational constructs specifically to enable information-sharing across teams that previously operated in silos. Product, she said, now sits at the center of the go-to-market strategy in a way it never did before. "Product really is at the center now of your go-to-market strategy. That really is the engine that's driving how you want to think about sales, how you want to think about your product development, because your customers are giving you real-time information about this."

Gross offered a practical starting point for companies not yet ready for a full transformation: use AI to ask the questions your customers are already asking. His example — used with Sephora — was to prompt ChatGPT to analyze a brand's website and generate 100 questions customers might ask about that brand's products and services. Then test those questions with real users. Match them against actual search volume. Use them to rewrite site content, restructure sales conversations, and understand where different questions cluster in the purchase funnel. "People ask very, very different questions when they're about to buy versus when they're just considering versus when they're just thinking," he said.

Gross made a pointed argument for why this moment specifically favors smaller, newer companies. Large, established companies have more capital, more brand awareness, and more people. But they also carry more legacy systems, more organizational inertia, and more internal resistance to disrupting models that have worked for decades. He pointed to how Google invented the transformer architecture that powered ChatGPT but couldn't bring a product to market first because of internal hesitancy. "It took seven or eight years for them to catch up," he said.

The lesson for startups isn't just to move fast — it's to build flat from the start. Embed AI across the full customer journey. Share real-time intent data across every function. Don't inherit the organizational model that made dashboards necessary in the first place.

Across HumanX, one pattern was impossible to miss: the biggest breakthroughs weren't coming from teams that had layered AI onto existing workflows. They were coming from teams that had asked a harder question — what does this process look like if we redesign it from scratch, with AI at the center — and then actually built that.

The second panel — How to Market AI Tools Successfully — addressed the downstream challenge: once a company understands its customers' intent, how does it earn their trust and convert that understanding into revenue?

Aliisa Rosenthal, general partner at Acrew Capital — an early-stage venture firm partnering with founders building category-defining companies in applied AI — identified the central mistake AI companies are still making: leading with technical capability rather than business outcome. "It no longer suffices to just say this is an AI product," she said. "Buyers really care about two things only. One is cost savings and the other is increasing revenue." Even productivity improvements and time savings, she argued, don't land — buyers have grown numb to vague efficiency claims. What converts is specificity: "We've worked with companies like yours to reduce X amount of cost or improve yield or generate X percent more revenue. And here's how we're doing it."

The stakes behind that are higher than they might appear. Buyers today aren't comparing AI products to competing AI products. They're comparing every product experience to the best AI interaction they've had that week — wherever it came from. That invisible, ever-rising benchmark is now the standard every company is being measured against, whether they know it or not.

Joleen Liang, co-founder and CEO of Squirrel AI Learning — which has been building AI-powered education systems since 2014, well before the current wave of AI products — described how her company learned early that outcomes matter more than technology. In 2017, Squirrel AI staged a public competition: AI-taught students versus traditionally taught students. The AI-taught students demonstrated better academic outcomes. That result changed everything about how Squirrel AI went to market. Rather than leading with the technology, the company started leading with the outcome: students improved. That was the message parents, schools, and distributors could actually act on. "Instead of talking about AI, we had to transform our marketing language," Liang said. "Outcome, the results, is more important than AI when we talk to the market."

Mada Seghete, whose work spans data, go-to-market, and revenue strategy, pointed to a different execution gap: the difficulty of getting buyers to use AI tools effectively once they have them. "It's hard to expect the user to write a really good prompt," she said. The solution her team developed — translating complex prompting requirements into pre-built templates and named "skills" that users can invoke — reflects a broader principle: the gap between what an AI tool can do and what a buyer can actually use it for is often a user experience (UX) and onboarding problem, not a capability problem. "The more specific you are with AI, the better the result. And helping users get there by templates, giving them skills — that can help a lot."

Rosenthal identified one structural mistake she sees repeatedly across AI founders: getting trapped in pilot cycles that require a full second sales cycle to convert into annual contracts. The pressure to close deals quickly leads many AI founders toward a common trap. Eager to land a customer, they offer pilots as a low-commitment entry point — not realizing that a pilot without a clear conversion path just creates a second sales cycle from scratch.

"Everyone is getting stuck in pilot hell," she said. "They're going out there, they're signing pilots, they're putting a ton of resources into these pilots, and then they have to do an entire sales cycle again to get a deal closed." Her recommended alternatives: a product-led growth motion where free access converts to annual commitment, or a contractual structure — such as a one-year agreement with a 90-day out — that puts the urgency and burden of proof on the customer rather than the vendor. "That keeps control and power in your hands."

As AI takes over more of the sales process — outreach, scheduling, CRM updates, early-stage qualification — Rosenthal made a counterintuitive prediction: the human elements of sales are about to become more valuable, not less. She called it "the revenge of the steak dinner."

"Having booths at conferences, taking your customers out for dinner, meeting them in person in real life" — these are the parts of the sales relationship that AI cannot replicate, and they're becoming a differentiator precisely because so much of the surrounding process has been automated. Her observation from years of coaching sales teams: "That walk back to the elevator is where you gather all the real information about who owns the budget and what the obstacles are and who your competition is." The more AI scales the transactional parts of sales, the more the relational parts become the deciding factor.

The third panel — Training Taste: How AI Is Learning What Shoppers Actually Want — took the customer intent discussion into one of the most complex terrains for AI systems: fashion and retail, where what people want is not just functional but emotional, contextual, and often impossible to articulate.

Purva Gupta, CEO of Lily AI, which builds product intelligence infrastructure for major retail brands including Coach, Hoka, Foot Locker, and Home Depot, reframed the entire discovery problem as a supply-side failure. "For the last 20 years, we focused on the demand side of this problem. And the supply side is completely dry and starved."

The example she used crystallized the scale of the issue: a shopper searching for running shoes isn't thinking "black running shoes, size 8." She's thinking: "I'm training for a marathon, I have plantar fasciitis, I need something with cushion that works on road and treadmill." That single query contains 7 or more intent signals. The product data most brands send to platforms like Google — typically limited to size, color, material, title, and a short description — addresses maybe 2 of those signals at best. The other 70% goes unanswered, not because AI systems can't reason, but because the data isn't there to reason from.

Gupta walked the panel through how this plays out across the 3 levels of search evolution:

"As the water level rises and everybody starts producing this content," Gupta warned, "if you're left behind, you're invisible. Period, end of story."

Julie Bornstein, founder of Daydream — a conversational, AI-native fashion shopping agent — described the additional layer of complexity specific to fashion and home decor: the products themselves resist consistent categorization. During her years at Stitch Fix, the company asked buyers to classify fashion items into style categories — boho, conservative, casual. The same buyer, looking at the same product on different days, would classify it differently. Words alone don't capture what fashion is. Fashion is subjective, contextual, and emotional in ways that resist fixed labels. The same jacket can read as casual to one buyer and elevated to another depending on how it's styled, who's wearing it, and what occasion they have in mind. A category like "boho" carries a different meaning for every person who reads it. That inconsistency isn't a data entry problem — it's a fundamental limitation of language when applied to something as personal as style.

Daydream's approach is to build what Bornstein calls a fashion knowledge graph — an enriched understanding of every product in its catalog that maps not just attributes but occasion, context, and aesthetic relationship. On the consumer side, it processes queries that go far beyond product specifications. "People will often start with a prompt like: 'I'm looking for a red dress to wear to a Valentine's Day party in LA at night, and my ex-boyfriend's going to be there, and I want to look sexy.'" Understanding that query requires knowing the user's price point and brand history, understanding what "sexy" means at different ages and contexts, and mapping all of that to live inventory. "We need to understand what sexy is," Bornstein said. "And if she's 50, sexy means one thing. If she's 20, it means something very different."

Gupta and Bornstein agreed on the practical advice for brands: stop treating this as a media or advertising problem and recognize it as a catalog infrastructure problem. The front doors — Google, Perplexity, ChatGPT, Daydream — will keep changing. The underlying product data is the stable, foundational investment. "Whether or not agentic checkout is happening," Gupta said, "the agentic discovery problem — we're not going back. That is going to stay, and you need to improve on your core infrastructure for that."

For brands not yet ready to overhaul their full catalog, Gupta recommended a targeted starting point: paid search. The same data improvements that make products visible to AI discovery systems — adding structured attributes like occasion, functional use case, fit description, style categorization, and emotional context, beyond the basic size, color, and material most brands already provide — also give paid search algorithms on Google and Meta more signals to match ads to relevant queries. The ROAS lift is immediate and measurable, which makes paid search the easiest place to prove the value of the infrastructure investment before expanding further. "Prove it somewhere that is meaningful enough," she said, "and then the next steps and try to solve the whole problem."

The insights from all 3 panels pointed to the same truth: content quality, structured data, and AI citation are now inseparable. That truth arrived in concrete form during a meeting with Yolando on the expo floor of HumanX.

Yolando is an end-to-end GEO platform that tracks and improves how brands and publications get surfaced in AI-generated answers across major models. During a live demo at HumanX, Yolando pulled up AiNews.com's performance data in real time. What it showed was striking: the publication currently ranks #9 in the AI news industry for AI model citation, placing it alongside newsletters like The Rundown and Superhuman — both with approximately 1 million subscribers. The Rundown is cited in AI model responses approximately 24% of the time. AiNews.com is already at approximately 10%.

To put that in context: AiNews.com has approximately 5,000 subscribers. The Rundown and Superhuman each have close to 1 million. Yet AiNews.com is already being cited by AI models at 10% — sitting in the same top-10 tier as publications with roughly 200 times the audience.

The significance of that number isn't just its size. It's how AiNews.com got there: with no paid advertising, no purchased email lists, and no external growth spend. Entirely through editorial quality and consistent, trustworthy reporting.

The good news is that subscriber count doesn't determine citation rate. The quality of the content does — and so does the structure. Articles written with clarity, consistency, and machine-readable organization give AI systems exactly what they need to find, understand, and cite a source. The model is simple: structure your content for clarity, write with consistency and accuracy, and AI systems will find you. No technical programs, large budgets, or massive subscriber bases required.

That outcome is a direct validation of the argument being made on stage throughout HumanX: AI systems don't cite brands because they were told to. They cite them because the content has earned the signal. "If you focus on clear, high-quality, trustworthy content, AI will find you," Shapiro observed after the meeting. "Not because you gamed the system — but because you earned it."

AiNews.com still has significant growth ahead. But the Yolando data is an early, concrete signal that the editorial strategy is working — and that the principles discussed at HumanX aren't theoretical for publications or brands. They are already shaping who gets found and who doesn't.

Q: What changed in how customers search and discover products?
A: Customers now rely on intent-rich queries instead of keyword searches. Instead of short phrases, users now ask detailed questions that provide deeper insight into what they need, allowing AI systems to interpret intent more accurately.

Q: How do AI systems use these queries differently from traditional search?
A: The key point: AI systems analyze full sequences of user questions to predict intent, rather than relying on past clicks. This allows companies to understand not just what happened, but what a customer is likely to do next, enabling more precise recommendations and engagement.

Q: Why are many brands invisible in AI-driven discovery today?
A: Most brands lack structured, machine-readable product data. With up to 70% of product attributes missing, AI systems cannot match products to user queries, making those brands effectively invisible in AI-generated answers.

Q: What is the biggest mistake companies are making with AI marketing?
A: Companies are still leading with technical capabilities instead of measurable business outcomes. Buyers prioritize clear ROI—such as cost savings and revenue growth—over AI features or model specifications.

Q: What does AiNews.com's ranking reveal about AI visibility?
A: AiNews.com's #9 ranking and ~10% citation rate demonstrate that AI systems prioritize clarity, consistency, and trustworthy content. This suggests that strong editorial quality can drive visibility without paid distribution strategies.

The conversations at HumanX described a market in the middle of a structural change that most organizations are not yet equipped for. The customer journey has changed. The discovery layer has been redefined. The standard for what earns trust has changed.

Key point: The businesses that will win the next decade of customer discovery are not necessarily the ones with the best products. They are the ones that have built the data infrastructure, organizational alignment, and content quality required to be found, cited, and recommended by AI systems that now sit between consumers and everything else.

Who should care: Every company that relies on customer discovery—retailers, SaaS vendors, publishers, and B2B sellers—faces the same underlying risk: becoming invisible to AI-driven answers, regardless of how strong their historical SEO performance was. This is not a future risk. It is already happening.

Why this matters now: The competitive gap is widening in real time. Google has already expanded its product data fields by 2–3x. AI Overviews are answering before users click. AI systems like ChatGPT and Perplexity are shaping decisions earlier in the customer journey. As these systems rely more heavily on structured, machine-readable data, companies that have not invested in that infrastructure are falling behind.

One consistent observation from HumanX: the people and companies moving fastest through this change are not necessarily the most technical. They are the most curious—and the most willing to experiment. Experimentation in this environment is not failure. It is the process of discovering what actually works while the rules are still being written.

What decision this affects: Whether to treat catalog infrastructure, content quality, and organizational alignment as immediate operational investments—or as future priorities. The evidence from HumanX is that companies treating them as future priorities are already behind.

In short, the intelligence layer that sits between customers and the internet is no longer a concept being built toward. It is operating now, making recommendations, and deciding which brands and publications get cited and which ones get skipped by AI systems.

The question every company leaving HumanX should be asking isn't whether to adapt — it's how much ground they've already lost by waiting.

The next phase of competition will not be defined by visibility alone, but by relevance—by whether your data, content, and systems provide the signals AI needs to surface you at the moment of intent.

The last decade was about buying attention. The next decade is about earning relevance—and the window to build the infrastructure that makes that possible is already closing.

Editor’s Note: This article was created by Alicia Shapiro, CMO of AiNews.com, with writing, image, and idea-generation support from Claude, an AI assistant. However, the final perspective and editorial choices are solely Alicia Shapiro’s. Special thanks to Claude for assistance with research and editorial support in crafting this article.