In the 2010s, CX was transformed by the smartphone and the apps it hosted. Brands rushed to ensure they had a “digital front door” that would allow customers to initiate multi-channel journeys at the tap of a screen. Today, those rules are being rewritten by AI-first customer journeys; demanding brands go back to the drawing board on everything from product listings to service processes.
When CX Network conduced its 2026 research into the state of CX, practitioners in the network were asked to select the three trends influencing their role the most from a list of more than 20 choices. AI-first customer journeys/ customers using AI for product research emerged as the third most selected response.
Whether facilitated by ChatGPT’s Instant Checkout, Mastercard Agent Pay or Google’s suite of agentic commerce tools, the AI-first customer journey can take many forms. Some key tools include:
Instant Checkout: Allowing customers to conversationally search for specific products, Instant Checkout gives customers the chance to buy their chosen products without leaving the ChatGPT interface.
Gemini and Google Search: Google’s agentic commerce suite allows customers to bypass the retailers they buy from, either by making their purchases in Gemini or AI Mode on Search, both of which draw on Google Shopping Graph, the AI-driven, real-time dataset containing over 50 billion product listings. Google’s agentic commerce suite also includes tools for retailers.
Perplexity: Promising to understand intent, remember preferences, and act as “an extension of how users would approach a task on their own”, Perplexity remembers the user’s past searches and learns patterns in known buying behaviors, removing the cold starts other agentic tools cannot avoid. It presents results in product cards that eliminate the endless scroll, with key product specs easily visible.
Agent Pay: Rolled out by Mastercard in 2025, this secure payment function is specifically designed to support agentic commerce. It enables AI agents to securely initiate and complete transactions using Mastercard’s advanced technologies, including network tokenization, authentication, and fraud and cybersecurity solutions.
Why is AI shopping popular?
AI shopping assistants are taking off because search friction is real and infinite choice is overwhelming – consumers no longer want to search multiple websites, or compare five products from a single vendor. They want machines to search the internet for them and return only the most relevant product choices, based on specific criteria shared.
The tech companies behind the first AI shopping assistants say their users have a higher intention to purchase and that retailers remain the “merchant of record”, allowing brands to maintain communications with customers, own returns, and build loyalty.
However, that does not mean retailers can passively wait for more sales, with no further action needed.
With AI shopping assistants assessing everything from delivery times and costs to service processes, organizations must ensure their product listings, company information, and customer policies – whether that’s returns information or contact details – are machine readable and competitive.
Joshua Curtis, customer care center manager for Super Retail Group, says:
“For practitioners, an AI-first journey forces a real shift in mindset. You’re no longer just managing channels or touchpoints you’re managing how confidently a system can guide a customer to the right outcome.”
He adds: “In an AI-first world, ambiguity becomes expensive. If pricing rules, delivery promises, or escalation paths aren’t clear, the journey either breaks or falls back to a human. That increases cost and usually frustrates the customer at the same time. So a lot of the work becomes about simplifying decisions, standardizing processes, and making things explainable.”
How buying agents have changed product discovery
The key change for brands is in how customers discover them.
“AI-first journeys change discoverability in a pretty fundamental way,” says Curtis. “It’s no longer just about how well you rank in search or how strong your marketing is, it’s about how well your business actually holds together when an AI is assessing it on a customer’s behalf.”
LLMs and agentic commerce platforms aren’t just looking at brand or price, Curtis adds. “They’re picking up on things like delivery reliability, how easy returns are, customer sentiment, and how often customers need to contact you to fix something. In effect, AI starts judging brands on operational reality, not just brand promise,” he explains.
“That’s a big shift. It means discoverability becomes tightly linked to service, fulfilment, and post-purchase experience. If delivery timeframes aren’t reliable, policies are hard to interpret, or customers regularly need help to resolve issues, AI will learn that quickly and steer customers elsewhere.”
From a CX point of view, Curtis says this “really lifts the importance of getting the basics right at scale”, including clean data, clear policies, and consistent experiences across channels.
Sam Davis, senior director of solutions engineering for the EMEA region at Yext says brands need to move away from the traditional “content as pages” mindset and adopting content as entities. “AI doesn’t think in pages, it understands structured data and linked relationships. Brands that provide structured, entity-based content give LLMs a trusted source to pull from, which directly impacts what gets surfaced in AI summaries,” he explains.
“As AI becomes the landing page for customer journeys, brands need to ensure their information is discoverable inside LLM experiences. Even with internal-facing use cases, brands are leveraging AI applications to aid retrieval of information across fragmented data sources via a federated search approach using an MCP (Model Context Protocol) server,” Davis says.
This new reality means reviews, listings, and self-service FAQs all need to be optimized for accuracy, kept up to date and clearly structured so AI systems can confidently understand and act on them.
Davis explains: “If an AI agent is booking a service, recommending a product or choosing between brands, it relies on structured, verifiable facts from a range of citation sources, such as web pages, reviews, listings, FAQs, help articles, etc. In what we're calling the 'answer economy' – a world where AI acts on brand data – traditional consumer touchpoints like websites may be required less for human interaction and more for AI. These digital assets are becoming the trusted data inputs that power agentic decision-making and commerce.”
What is AEO and GEO?
As explained by Curtis and Davis, discoverability is no longer rooted in traditional advertising, marketing, and SEO. Today it’s all about GEO (generative engine optimization) and AEO (answer engine optimization) and this calls for site-wide transformation and optimization.
AEO focuses on providing direct, concise answers for featured snippets and voice search, while GEO focuses on being cited as a trusted source within broader AI-generated summaries.
Calling AEO one of the “most transformative shifts” in CX, Ekaterina Mamonova, global head of broker proposition for Liberty Specialty Markets, says: “With generative AI becoming the primary research tool for both consumers and B2B buyers, brands must ensure their expertise is surfaced accurately and authoritatively by AI assistants. High‑quality, question‑led content, structured for machine readability is fast becoming a critical acquisition lever. Organizations must embrace AEO to support the shifts in discoverability and capture the customers who use generative AI as their primary research tool.”
“Buyers now expect hyper‑personalized journeys, authentic interactions, and seamless digital discovery, all while privacy regulations tighten and third‑party data disappears,” she says. “The organizations winning new customers today are those that reimagine acquisition not as a volume game, but as a value‑building discipline rooted in relevance, trust, and frictionless engagement.”
Meanwhile, GEO is dependent on structure, conversational relevance, and trust.
Davis explains: “Success in generative search is not driven by keywords alone, but by the quality, structure and authority of a brand’s underlying information.”
Structure: AI desires structured data and a clear recognition of appropriate and relevant linked-entity relationships.
Conversational relevance: Search queries are becoming longer, more conversational and session based. “Brands need content that can answer an initial question and continue supporting follow-up questions with consistent, accurate responses within the same interaction,” Davis says.
Trust: “This is essential,” says Davis. “Consistent, structured data distributed across multiple consumer touchpoints could contribute to LLM training, significantly improving how generative AI systems retrieve, trust, and include your brand in answers.” Disclosure is also crucial. If AI helped write branded content or updated some information, brands must make this clear. “Who wrote it, when it was updated; stating whether AI was involved should become the norm to give consumers transparency,” Davis adds.
The future of agentic commerce and online shopping
In these nascent stages, the AI shopping game is anybody’s to win.
Recent advances from Big Tech have levelled the playing field and, in theory, smaller players with the agility to act on emerging trends can take advantage of the current situation to gain significant market share and prominence.
The arrival of AI shopping assistants and tools from the likes of Google, OpenAI, and Perplexity AI also threaten the online shopping incumbents, such as Amazon and eBay, both of which have already demonstrated a love/hate relationship with third-party shopping assistants.
Data published in 2025 by Yext found that 40 percent of consumers in the UK use AI search at least once a day, while 80 percent had increased their use of AI search over the past 12 months.
“With this growing confidence in AI, we’re also seeing rising customer expectations for fast, intelligent answers,” says Davis. “As this space continues to evolve, we’ll see further shifts in customer AI search journeys as AI moves from offering recommended actions to agentic “do it for me” processes. Over time, AI search becomes less about discovery and more about completing tasks efficiently, which reshapes the end-to-end customer experience entirely.”
These changes mean discoverability must be treated “like infrastructure”, Davis says. “Ultimately, AI visibility begins with structured data, so brands who prepare their ecosystems to ensure they show up consistently in AI search will gain an early advantage. In today’s answer economy, consistency builds trust and trust builds presence,” he adds.
Agentic commerce and customer loyalty
Of course, when customers are enabled to bypass the retailers they buy from, questions are raised about what this means for loyalty. While loyalty and points programs must be optimized – and sometimes simplified – for machine customers to compare, loyalty itself must also adapt.
Writing for CX Network, Anjali Jain, AI and enterprise architect and a senior tutor in AI and Machine Learning at the University of Oxford, says that in an AI-first world, practitioners must take four steps to strengthen loyalty:
- Invest in strong first party data capabilities to maintain direct understanding of customer behaviour.
- Redesign loyalty programs around effort reduction, clarity, and continuous value.
- Use AI to provide timely, context aware support throughout the customer journey.
- Create physical and digital experiences that deliver emotional connection and reinforce the brand
The changing role of the contact center
The arrival of agentic commerce also changes the role of the contact center. With fewer contacts, interactions are now based on quality, rather than speed.
“As more intent capture and resolution happens upstream, service teams spend less time answering ‘where is my order?’ and more time handling exceptions, recovery, and emotionally charged situations,” Curtis says. “The volume may reduce, but the importance of those interactions increases.”
He adds: “For CX leaders, that means designing journeys with failure in mind. Where can AI confidently resolve? Where should it step back? And how do we make the handover feel seamless for both the customer and the agent? In practice, success comes down to data quality, strong knowledge foundations, and tight alignment between digital, service, and fulfilment teams. AI-first journeys expose gaps very quickly, so operating in silos just doesn’t work.”
As much as this holds true for now, AI moves fast and soon, LLM-powered commerce will evolve and improve, paving the way for next innovation.
With an eye on how the technology is likely to develop, Dianna Lyngholm, director of website and creative services at FUN.com & HalloweenCostumes.com, says: “Answering six questions in order to get to the product that you're looking for isn’t a great experience. I think short-term, it can definitely raise some revenue, but I think in the next three years, LLMs probably won't be as popular as they are today.
“I would say the next step is vector AI. LLM have no long-term memory. You go into agentic AI and 15 questions later, you're finally at what you want. Vector AI solves the cold start issue. Instead of having to answer a bunch of questions when you're on a website, you start clicking and searching and, and everything gets personalized to you in the moment, no latency there. I think that's really the way it's going to go,” Lyngholm adds.
Quick links
- When your customer is a machine: Rethinking service design for AI agents
- Agentic Agility: The new ‘normal’ for CX in 2026
- The human-machine translator: CX's most important new role