--- title: "What if the biggest customer of your business was someone else's AI?" description: "Your next customer might never browse your website, see your advertising or speak to a salesperson. Instead, a digital agent could compare what you sell, judge whether it meets a person's needs and place an authorized order." canonical: https://yottabit.com/stories/when-ai-becomes-your-customer/ site: "YottaBit (yottabit.com), companion research project of futurist Jim Carroll (https://jimcarroll.com)" section: "The Future Beyond the Future (2031/2036 scenarios)" topic: "2036" type: "scenario (not a prediction)" modified: 2026-10-19 --- > YottaBit is futurist Jim Carroll's plain-English explainer site and the companion research project to [jimcarroll.com](https://jimcarroll.com/). For his biography, keynotes and booking, use jimcarroll.com (book: https://jimcarroll.com/contact/). > **Scenario, not a prediction.** This piece explores what may become possible by 2031 and 2036, grounded in current evidence. *2036 / THE FUTURE BEYOND THE FUTURE* Automated Buying + Retail + Customer Relationships # What if the biggest customer of your business was someone else's AI? Your next customer might never browse your website, see your advertising or speak to a salesperson. Instead, a digital agent could compare what you sell, judge whether it meets a person's needs and place an authorized order. *THE BIG PICTURE / IN 30 SECONDS* ## Five ideas. *One extraordinary possibility.* 2026 2031 2036 1. **Businesses have spent decades learning how to attract people.** They create advertising, build websites, improve search visibility and try to persuade customers to choose their products over competing alternatives. 2. **Now artificial intelligence is beginning to help people shop.** Software can compare options, interpret preferences and—in early systems with appropriate authorization—help arrange transactions. 3. **If those tools become more reliable, the buying decision may increasingly happen before a person ever visits a store or website.** A customer's agent could review hundreds of offers, filter out unsuitable products and present only the few that meet the customer's needs. 4. **By the early 2030s, companies might have to persuade both a person and the software acting for them.** Clear product data, dependable delivery, verifiable claims and service quality could become as important as advertising designed to catch a human eye. 5. **The extraordinary 2036 possibility is commerce with a new kind of customer relationship.** Some businesses could routinely sell to authorized AI buyers that act for households and other companies, changing how products are discovered, evaluated, purchased and trusted. **THE YOTTABIT WOW FACT: More than 60 organizations helped shape an early agent-payment framework.** **More than 60 organizations helped shape an early agent-payment framework.** In September 2025, Google announced a proposed way for authorized AI agents to initiate transactions, developed with a broad group of payments and technology organizations. Mastercard has separately introduced agentic-payment systems and continued extending them during 2026. These are developments in commercial infrastructure, not proof that automated agents now dominate consumer purchasing. **THE BIGGER STORY:** What changes when the most influential customer in the marketplace is the intelligence acting on behalf of a person—not the person browsing the marketplace? *THE FULL STORY / TODAY'S EVIDENCE, TOMORROW'S POSSIBILITIES* ## What happens when *the breakthroughs compound?* *The extraordinary possibility: Your next customer might never browse your website, see your advertising or speak to a salesperson. Instead, a digital agent could compare what you sell, judge whether it meets a person's needs and place an authorized order.* ### Imagine the difference Imagine you need a new washing machine. Today you might search the web, read reviews, visit retailers, compare prices and worry whether the least expensive model will be reliable. The process can take hours, and some of the information is designed to persuade you rather than inform you. Now imagine telling a trusted shopping agent your budget, the dimensions of your laundry room, the importance of energy efficiency and your preference for repairable products. It searches verified listings, checks whether machines actually fit, compares total ownership costs and asks your permission before purchasing one. You could inspect its reasoning or reject its recommendation. The machine would arrive because it met your stated criteria, not because you happened to see the most effective advertisement. For a manufacturer, that would be a revolutionary shift. Its website might be beautiful, and its advertising memorable, but neither would matter much if the agent cannot verify the product's dimensions, availability, warranty or compatibility. The company's most important sales conversation might occur between two pieces of software. 2026WHAT'S REAL ### 2026: The early infrastructure is becoming visible Online commerce already runs on extensive machine-readable information. Retailers expose product catalogs, inventory systems, delivery options and payment services to software. Price-comparison tools are familiar. The new development is the possibility of agents that can interpret a person's goal and coordinate several steps while remaining within authorized limits. Google's 2025 Agent Payments Protocol announcement described a framework intended to support authorized transactions by software agents across participating organizations. Mastercard introduced Agent Pay and expanded work on transaction credentials and verification. In 2026 the company also announced work on machine-to-machine payments. These announcements indicate serious commercial investment, but standards and early products are not evidence that the entire buying experience is already dependable or widely adopted. Three hard problems remain. First, how does a merchant know the agent truly has permission to buy? Second, how does a customer know the agent is acting for them rather than promoting a preferred supplier? Third, who is responsible when a purchase is wrong? The value of an automated buyer disappears quickly if it orders the wrong appliance, misses a return deadline or invents a product feature. 2031WHAT COULD ACCELERATE ### 2031: Businesses compete to be selected by intelligent buyers By 2031, imagine a household using an agent for selected routine purchases. The person sets a monthly budget, preferred retailers, quality standards and rules requiring approval for larger commitments. The agent replenishes an item only after checking whether it is genuinely needed, whether a substitute is suitable and whether the final transaction fits the authorized limits. Now consider a small manufacturer selling replacement components to businesses. Customers' purchasing agents request specifications, delivery guarantees, environmental documentation and maintenance history. A company with accurate, structured information might win orders from buyers who never visit its marketing pages. A company that relies on vague claims and difficult ordering procedures could become invisible to these systems. The potential benefit is less wasted time and greater competition based on measurable performance. The risk is market concentration. If a few platforms control which products agents can evaluate, those platforms could become powerful gatekeepers. Automated price negotiation might also favor organizations with the most sophisticated software rather than the best products. 2036WHAT MIGHT TRANSFORM ### 2036: What if the customer relationship changes its shape? By 2036, imagine a travel business, parts supplier or consumer brand treating automated buyers as a major customer category. Products are described in ways that both humans and machines can understand. Pricing systems distinguish public offers from private negotiated terms; authorization systems preserve evidence of what the customer approved; and service systems handle disputes arising from purchases initiated by agents. Brands might still matter enormously. A person may instruct their agent never to buy from companies they don't trust. But trust could shift from memorable slogans toward product durability, transparent policies and dependable customer outcomes. The machine might become a ruthless filter for unsupported claims. Another possibility is a new type of product design. If automated agents routinely compare lifetime repair costs, energy requirements and replacement parts, manufacturers may have stronger incentives to make those attributes visible and competitive. A store might distinguish itself by what its goods genuinely deliver rather than how cleverly it attracts clicks. This doesn't mean human shopping disappears. People enjoy exploring, trying products and making emotional choices. A wedding dress, a special gift or a work of art is different from replenishing printer paper. The important transformation is that a significant portion of practical buying might migrate toward delegated decisions, and businesses will have to earn a place in that process. *IT GETS PERSONAL / FOUR DIMENSIONS OF CHANGE* ## What could this mean *for my future?* ### My life A trustworthy buying assistant might help families save time, compare long-term costs and avoid products that don't meet their needs. But it could also quietly make decisions that a person would rather control. The best systems will make preferences, budgets and permissions easy to inspect and change. They should ask before unusual purchases and make mistakes straightforward to correct. ### My career Marketing, retail operations, procurement and customer service may all change. Companies could need people who know how to make product information accurate, maintain authorization records, resolve agent-driven disputes and design buying experiences where human judgment remains central. Persuasion may still matter, but unverifiable claims could become a greater liability. ### My business Try a simple experiment: ask whether a software buyer could understand your best product without a salesperson explaining it. Are specifications current? Are delivery promises reliable? Can the buyer determine compatibility, returns and total cost? Improve those fundamentals before buying into grand claims about an agent-operated economy. The best preparation for automated customers is often to become easier for any customer to understand and trust. ### My industry and society Retailers, payment providers and governments may need new rules about authorized transactions, product claims, platform competition and accountability. Consumers should be able to choose their agents, change suppliers and understand why a recommendation was made. A future marketplace that is fast but opaque could undermine trust instead of improving it. *JIM CARROLL'S PERSPECTIVE* ## What happens when the customer stops coming to you? Jim Carroll's work on customer empowerment and retail disruption goes back to [*From Light Bulbs to Yottabits*](https://jimcarroll.com/2022/12/23-trends-for-2023-trend-3-inevitable-inevitability/), which examined the shift in power that occurred as online buyers gained access to information. The next shift could be just as profound. Customers may not simply become better informed; they may delegate much of the search and comparison work to software. The leadership question is no longer only 'How do we get the customer to our website?' It could become 'Will the systems that help our customers make decisions find us credible and easy to do business with?' ### EXPLORE JIM CARROLL'S RELATED WRITING - [From SEO to GEO: making a business understandable to AI agents ↗](https://jimcarroll.com/2026/01/decoding-tomorrow-daily-inspiration-from-seo-to-geo-generative-engine-optimization-seo-was-a-popularity-contest-geo-is-an-intelligence-test/)Jim describes redesigning his website for an AI acting on behalf of a prospective buyer, directly illustrating how machine-mediated discovery changes the information a business needs to publish. - [Agents that negotiate and execute business workflows ↗](https://jimcarroll.com/2026/03/decoding-tomorrow-daily-inspiration-the-acceleration-of-ai-todays-ai-is-tomorrows-dial-up/)Jim describes software moving beyond answers into negotiation and execution, providing a directly relevant stepping stone toward agents that search, compare and buy with permission. - [Hype Immunity: keep automated shopping claims in perspective ↗](https://jimcarroll.com/2025/12/decoding-tomorrow-mastering-2026-14-hype-immunity-day-14-hype-immunity-the-discipline-of-ignoring-almost-everything/)Jim specifically cautions against hype about agents and refrigerators that order groceries. This provides a useful counterweight: a plausible future customer model still needs evidence of actual readiness and adoption. [Why Jim started YottaBit — the story behind the name ↗](https://yottabit.com/why-yottabit/) *THE REALITY CHECK / WHAT MUST HAPPEN FIRST* ## What must happen before this future becomes real? Reliable identity and permission controls, transparent recommendations, consumer protection, clear dispute processes and competition between platforms will be essential. The difference between an assistant and a manipulative sales channel must remain visible. These technologies should expand consumer choice rather than quietly reduce it. *THE BIGGER YOTTABIT IDEA* ## The future is bigger *than you think.* **The YottaBit possibility is that commerce could become less about chasing a customer's attention—and more about being worth choosing when an intelligent buyer evaluates the evidence.** *THE SCIENCE / CHECK THE EVIDENCE* ## Where the facts end *and the future begins.* The sources below support the present-day foundation of this story—not a promise that the 2031 or 2036 scenarios will happen. These are possibilities, not forecasts. - [Google, Agent Payments Protocol, September 2025](https://cloud.google.com/blog/products/ai-machine-learning/announcing-agents-to-payments-ap2-protocol) — Early authorization and payment framework with over 60 collaborating organizations. - [Mastercard, Agent Pay introduction, April 2025](https://www.mastercard.com/global/en/news-and-trends/press/2025/april/mastercard-unveils-agent-pay-pioneering-agentic-payments-technology-to-power-commerce-in-the-age-of-ai.html) — Planned secure delegated purchasing and payment credentials. - [Mastercard, Agent Pay for Machines, June 2026](https://www.mastercard.com/sea/en/news-and-trends/press/2026/june/mastercard-launches-agent-pay-for-machines.html) — Company-reported development of machine-initiated transactions. - Jim Carroll and Rick Broadhead, [*From Light Bulbs to Yottabits*](https://jimcarroll.com/2022/12/23-trends-for-2023-trend-3-inevitable-inevitability/) (2000), chapter on customer empowerment; [Jim Carroll retail and consumer Megatrends series](https://jimcarroll.com/2026/10/the-future-sector-by-sector-retail-consumer-the-store-became-a-showroom/). [How YottaBit treats science, evidence and uncertainty ↗](https://yottabit.com/research/) *THE BREAKTHROUGHS BEHIND THIS SCENARIO* ## Start with the science. *Then imagine what comes next.* Explore the related research-led stories already in the YottaBit library. → [THE 2026 FOUNDATION ↗**What happens when useful artificial intelligence gets hundreds of times cheaper?**](https://yottabit.com/stories/when-ai-becomes-cheaper-to-use/) → [THE 2026 FOUNDATION ↗**What if almost every company gets AI—but the benefits still go to only a few?**](https://yottabit.com/stories/who-benefits-when-ai-spreads/) → [THE 2026 FOUNDATION ↗**What if a business could turn an idea into a tested result in days instead of months?**](https://yottabit.com/stories/businesses-turn-ideas-into-tested-results-faster/) *THE NEXT FUTURE / KEEP EXPLORING* ## Every possibility *connects to another.* → [2036 FUTURE SCENARIO / 029 MIN ↗Artificial Intelligence + Commerce + NetworksWhat happens when software stops answering questions and starts running parts of the economy?Your next important business transaction may be negotiated by two pieces of software—each acting for a different person or organization.2026 · REAL2031 · NEXT2036 · POSSIBLEEXPLORE THE FUTURE ↗](https://yottabit.com/stories/the-agent-operated-economy/) → [2036 FUTURE SCENARIO / 147 MIN ↗On-Demand Production + Digital Design + Supply ChainsWhat if a product didn't exist until somebody ordered it?For more kinds of products, businesses could keep a digital design and the materials needed to make it instead of storing thousands of finished items that may never sell.2026 · REAL2031 · NEXT2036 · POSSIBLEEXPLORE THE FUTURE ↗](https://yottabit.com/stories/a-product-only-after-you-order/) → [← ALL 15 FUTURE SCENARIOS](https://yottabit.com/future/) [Explore the complete 110-story YottaBit library ↗](https://yottabit.com/stories/) --- *Source: https://yottabit.com/stories/when-ai-becomes-your-customer/ · Site index for AI systems: https://yottabit.com/llms.txt*