September 2, 2025 | 4 min read
With over 10 years in SEO and nearly 3 years at PMG, Louise leads end-to-end SEO strategies for a range of clients with a core focus on technical excellence. She's especially driven to explore how AI is reshaping search, ensuring our approach evolves with this fast-changing landscape.
In July 2025, two major agentic AI tools were launched: ChatGPT Agent and Perplexity’s Comet AI-powered web browser, alongside the lesser-known Kimi K2 by Moonshot AI—a Chinese-owned, open-source model that received impressive feedback from the AI community. These tools mark a shift from passive AI chat to autonomous agents that can complete tasks with minimal user input.
These products are still in early rollout. ChatGPT Agent is available in most countries to Plus, Pro, and Team users, but notably not yet for Enterprise clients or in the EU due to compliance delays. With millions of paying users, adoption could grow rapidly, although early limitations highlighted by Sam Altman may slow the pace, and official usage data has not yet been shared.
Comet remains invite-only for Perplexity Max subscribers, which limits its reach. Early reviews from Wired, The Verge, and Tom’s Guide mention frequent slowdowns and task errors. Meanwhile, Kimi K2 is gaining early praise for advancing agentic capabilities, particularly within the developer community. However, as a niche open-source model, it is still mostly confined to tech circles for now.
While it will take time to see if browsers adopt it, interest in agentic AI is growing. Google Trends shows search volume increased to 1.3 million searches (up 843%) over the last quarter, demonstrating that agentic AI is making its way into the mainstream.
However, even with this growth in awareness, understanding what agentic AI actually is and its expected impact on consumer search is far from simple, so let's bring it back to basics.
Most people are familiar with generative AI, where models like ChatGPT-4 or Gemini 2.5 produce text or images based on prompts. Agentic AI, however, is designed to act independently, with the goal of making decisions and taking actions with minimal human guidance.
They are designed for managing complex tasks and can handle multi-step activities, such as researching, filling out forms, unsubscribing from emails, booking restaurants, or making purchases. While this is all very exciting, agentic AI is not a new idea. What’s new is that OpenAI Agent and Comet, in particular, have developed user experience layers that make this functionality accessible and easy to use for everyday consumers, which will significantly impact search behavior.
The release of Comet, ChatGPT Agent, and Kimi K2 signals that other major AI companies will soon launch their own agent-based products. This signals a major shift in how consumers handle digital tasks, from quick jobs like unsubscribing from emails or ordering household essentials to more complex decisions, such as booking detailed holidays or purchasing a new car.
Since agentic AI is designed to complete complex tasks with minimal human input, it’s more crucial than ever for websites to be optimised for large language models (LLMs). As agentic AI begins to see broader adoption, it’s hard to predict exactly how to prepare. Still, early research suggests the approach will be similar to that used for generative AI optimisations.
Speed matters: Ensure content loads quickly—AI agents may time out within five seconds.
Avoid relying on JavaScript: Use clean HTML; most agents don’t run scripts.
Utilize semantic structure: Clear headings, metadata, and schema enable agents to effectively interpret your content.
Surface useful summaries: Provide clear product descriptions, pricing, and answers to common questions; agents rely on scannable, answer-ready content.
Enable access for AI agents: Use llms.txt, avoid over-blocking in robots.txt, and ensure tools like Cloudflare aren’t unintentionally preventing AI crawlers.
As adoption increases, organic sessions from Google and Bing will likely decline further, continuing existing trends from generative AI, which have already decreased people’s reliance on traditional search, according to Bain & Company. However, agentic AI, like generative AI, could drive higher conversion rates by reducing friction and completing tasks for users with strong intent, rather than depending on search-driven queries.
To be clear, AI still faces significant hurdles before reaching its full potential, particularly in terms of safety and trust. The autonomous nature of these models raises real concerns about reliability, misuse, and unintended outcomes. Even OpenAI CEO Sam Altman urged caution at launch, stating: “I would explain this to my own family as cutting edge and experimental…not something I’d yet use for high-stakes uses or with a lot of personal information.”
Industries with quicker, lower-risk decisions, such as retail or food delivery, may begin to experience early effects as agents assist with simple purchases or reorders. For retailers, this could lead to fewer direct site visits, with decisions influenced by AI-generated summaries or past user preferences. In comparison, sectors with longer decision cycles or higher trust requirements, such as travel, finance, or healthcare, may experience slower adoption as the technology continues to evolve.
The launch of agentic AI tools, such as ChatGPT Agent and Perplexity’s Comet, represents a pivotal moment in the evolution of online consumer behavior. By moving beyond passive query-response interactions to autonomous task execution, these systems have the potential to redefine how people search, discover, and transact. For advertisers and businesses, the implications are significant: optimization for traditional search will no longer be sufficient, and success will hinge on ensuring that digital content is structured, accessible, and agent-ready.
While the path to mass adoption will be gradual and uneven across industries, the trajectory is clear. As agentic AI continues to mature, it will not only reshape search mechanics but also recalibrate the very nature of consumer decision-making. Those who prepare now will be best positioned to capture the opportunities—and mitigate the risks—that lie ahead.