AI Search Is Moving Beyond Answers
Most businesses are still trying to understand AI search as a question-and-answer problem. They want to know whether they appear in ChatGPT, whether Google AI Overviews mention them, whether Perplexity cites their website and whether buyers are seeing competitors more often.
Those questions still matter, but the next stage of AI search is already becoming clear. AI systems are not only answering questions anymore. They are starting to act.
That is why the Amazon vs Perplexity dispute matters. It is not just a legal fight between a large ecommerce platform and an AI company. It is an early example of a much bigger shift in how people may use the internet, where AI browsers and agents do not simply provide information, but browse, compare and complete tasks on behalf of users.
Why the Amazon vs Perplexity Case Matters
Perplexity’s Comet browser and associated AI agent were designed to browse and complete tasks on behalf of users. Amazon accused Perplexity of accessing private customer accounts through that browser and agent, and of disguising automated activity as human browsing. Perplexity has argued that users should have the right to choose their own AI tools.
The case has now reached a US appeals court, where judges are considering how a computer fraud law from 1986 applies to modern AI agents. That is what makes the story so interesting. The law was written decades before AI agents could log into websites, interact with accounts, place orders and act with limited human oversight.
The dispute raises a much bigger question: when an AI agent acts online, who is really acting? Is it the user, the AI company, the browser, or an automated system accessing a platform without permission? The answer could shape the next phase of search, discovery and online commerce.
The Bigger Commercial Question
At first glance, this looks like a shopping dispute. Amazon does not want Perplexity’s agent accessing its platform in a way Amazon says is unauthorised. Perplexity says users should be able to use AI tools to help them shop.
Underneath that is a more important commercial issue: who controls the customer journey when AI agents start acting for users? In the traditional internet model, platforms control much of the experience. Amazon controls the Amazon shopping journey. Google controls much of the search journey. Booking platforms, marketplaces and comparison sites shape how users discover, compare and purchase.
AI agents challenge that model. If a user can ask an AI browser to compare products, check reviews, find better prices, log into an account and place an order, the platform may lose some control over the journey. The AI agent becomes the interface.
Search Is Becoming Agentic
Agentic search is different from traditional search. Traditional search helps a user find information. The user still does most of the work. They search, click, read, compare, decide and act.
Agentic search starts to take on parts of that process. A user might ask an AI agent to find the best option, compare providers, assess reviews, check pricing, summarise terms, fill in forms, book a meeting or complete a purchase.
That changes the meaning of visibility. In traditional SEO, visibility often means ranking in search results. In AI search, visibility means appearing in generated answers. In agentic search, visibility may mean being selected by an AI system that is actively trying to complete a task.
What This Means for Businesses
Businesses need to understand that AI agents will judge websites differently from humans. A human visitor may respond to design, tone, brand personality, imagery and emotional cues. Those still matter, because humans remain part of the decision. But an AI agent is more likely to rely on structured signals, clear information, consistent claims and accessible evidence.
It will need to understand what the business does, who the business serves, whether the business is credible and whether it can find relevant service pages, pricing information, proof points, reviews, case studies, contact options and trust signals.
If that information is vague, inconsistent or hidden, the agent may struggle to select the business. This is why agentic search pushes businesses beyond traditional website design. A website cannot only look good to a human. It also needs to make sense to machines.
The Buyer Journey Could Become Much Shorter
Agentic search could compress the buyer journey. Instead of moving through awareness, consideration, comparison and enquiry across multiple sessions, a user may ask one detailed prompt and receive a narrowed shortlist.
For example, a buyer might ask: “Find me three UK-based data protection consultancies that work with SaaS companies, have experience with breach response, provide staff training and have strong public proof of credibility.”
In a traditional journey, that buyer might search Google, open several tabs, read websites, check reviews, ask LinkedIn, compare service pages and then shortlist providers. In an agentic journey, the AI system may do much of that comparison first. That means businesses could be filtered before they ever get a chance to pitch.
Trust Signals Will Matter More
Agentic search makes trust signals more important. When AI systems are only answering general questions, they can summarise broad information. But when they are helping users decide or act, the quality of the evidence becomes more important.
An agent recommending a provider, booking a service or helping with a purchase needs to rely on signals that suggest credibility. That could include client reviews, case studies, industry mentions, media coverage, directory profiles, accreditations, awards, founder credibility, clear policies, transparent contact details and consistent service descriptions.
For B2B and professional services businesses, this is especially important. Many credible companies have strong client relationships, referrals and expertise, but weak public evidence online. That may have worked in a referral-led market. It may not work as well in an AI-mediated market.
The Legal Debate Reveals the Commercial Direction
The Amazon vs Perplexity case is important because it shows that AI agents are already pushing against the boundaries of the current web. Websites were not designed for autonomous agents acting on behalf of users. Platform terms were not always written with AI browsers in mind. Security controls were not built for a world where a user may delegate browsing, comparison and purchasing to an AI system.
Platforms will want to control access, protect data, maintain security and preserve their commercial models. AI companies will argue that users should be able to delegate tasks to their chosen assistants. Businesses will need to decide how they want AI agents to interact with their websites, content and services.
This is not just a legal debate. It is a debate about the future architecture of the internet. If AI agents become common, websites may receive more machine-driven visits, brands may be evaluated before humans arrive, and content may need to serve both human readers and AI interpreters.
Agentic Search Changes GEO
This also changes how businesses should think about GEO. Generative Engine Optimisation has often been described as the process of improving visibility in AI-generated answers. That is a useful starting point, but agentic search expands the challenge.
GEO is not only about being mentioned. It is about being understood, trusted and selected. It is about making sure AI systems can interpret your business accurately enough to include it in answers, comparisons, shortlists and task-based recommendations.
As AI agents become more capable, businesses will need to optimise for the tasks buyers want to complete, not just the questions they ask. A GEO strategy should not only consider prompts such as “best cybersecurity company in the UK.” It should also consider task-based prompts such as “find three cybersecurity providers for a mid-sized financial services business and compare their incident response expertise.”
Machine-Readable Trust Is Becoming a Competitive Advantage
One of the biggest lessons from the rise of AI agents is that businesses need to make trust machine-readable. This does not mean reducing trust to a simple score. It means making trust evidence easy for AI systems to find, interpret and connect.
A case study should clearly explain the problem, the client type, the solution and the outcome. A service page should clearly explain who the service is for, what it includes and when it is relevant. A review profile should be active and credible. A founder profile should make expertise clear. Third-party mentions should reinforce the same positioning.
If these signals are scattered, inconsistent or hidden, AI systems may struggle to build confidence. Humans can sometimes fill in the gaps through conversation, referrals or brand familiarity. AI agents will rely heavily on what they can access and interpret.
What Businesses Should Do Now
Businesses do not need to rebuild everything immediately, but they should start preparing for agentic discovery. The first step is to audit how clearly their website explains what they do. If a human needs several pages to understand the company’s offer, an AI system may also struggle.
The second step is to review trust signals. Businesses should look at whether they have public reviews, case studies, media mentions, directory profiles, founder credibility and relevant third-party validation. The third step is to test task-based prompts, not just simple recommendation prompts.
Instead of only asking whether the brand appears for “best provider” queries, businesses should test more realistic agentic prompts that include buyer context, comparison criteria, location, service needs and decision factors. This will show whether AI systems can understand, compare and select the business in a real buyer scenario.
The Lesson for Businesses
The Amazon vs Perplexity case is not just about one AI browser shopping on one platform. It is a signal of where the internet is heading. AI systems are moving from answering questions to completing tasks. Search is moving from links to answers, and from answers to actions.
That changes what visibility means. It is no longer enough to rank. It is not even enough to be mentioned. Businesses need to be understandable, credible and selectable by AI systems acting on behalf of users.
The companies that prepare early will have an advantage. They will make their information clearer, strengthen their evidence layer, monitor how AI systems interpret them and think about how buyers use agents, not just how they use search engines. The AI browser has gone to court because the next version of the web is already arriving. Businesses now need to ask whether they are ready to be found, understood and chosen by the agents that may soon sit between them and their buyers.