The Agentic Buyer: How AI Is Starting to Make Purchasing Decisions

Something is happening that most brands are not paying attention to. AI is no longer just helping people find things. It is starting to choose things for them.

This is not a 2035 prediction. It is a description of behavior that is measurable today, accelerating, and has a specific name: agentic commerce.

What Agentic Commerce Actually Is

Agentic commerce describes AI systems that take autonomous research and purchasing actions on behalf of users. Instead of a person searching, comparing, and clicking, an agent handles the workflow end to end.

Early versions are already running in the wild:

• Travel agents that compare flights, hotels, and itineraries autonomously and book against stated preferences
• Replenishment systems that track household inventory and reorder preferred products without prompting
• Procurement agents that shortlist vendors, compare pricing, and draft purchase orders against requirements
• Restaurant assistants that suggest and book reservations based on occasion, dietary needs, and location

McKinsey's 2024 analysis of generative AI estimates the technology could add the equivalent of trillions of dollars in annual value across industries, with autonomous agent workflows representing one of the fastest-growing categories. Whatever the precise figure, the direction is unambiguous. Agentic decisions are starting to move real money.

The Invisible Filter

Here is the problem. When an AI agent decides what to recommend or buy, it draws from what it knows. It does not browse. It does not discover brands through a beautifully produced Instagram post or a well-placed billboard. It works from structured knowledge: information it encountered during training, retrieved through web search, or pulled from databases it has been configured to trust.

If your brand is not clearly represented in those sources, the agent does not know about you in the way it needs to in order to make a confident recommendation.

The result is not that the agent declines to recommend you. The result is that it recommends someone else, confidently, every time.

This Is a Different Problem Than SEO

Traditional search optimization is built around a human in the loop. A person searches. They scan results. They evaluate. They click. Even if your brand is not in position one, you can still catch the eye of someone scrolling the page.

Agentic commerce removes the human from that loop. The agent evaluates, selects, and acts. There is no results page. There is no second chance. The decision happens inside a process the consumer never sees.

The brands positioned to win in agentic commerce are not necessarily the ones with the largest ad budgets or the most followers. They are the ones whose positioning is legible to AI systems: structured data, clear entity definitions, consistent information across sources, and content that directly answers the queries agents are optimizing for.

What Agents Are Actually Looking For

When an AI agent evaluates a brand, hotel, restaurant, or product, it synthesizes signals from multiple sources. The factors that most consistently drive agent recommendations look like this:

1. 1.Entity clarity — Does the AI clearly understand what this brand is, what it offers, and where it operates? Ambiguity is disqualifying.
2. 2.Cross-source consistency — Do the brand's description, menu, pricing, and positioning match across its website, reviews, press coverage, and directory listings? Inconsistency creates low-confidence recommendations.
3. 3.Schema markup — Does the website communicate structured data that an agent can parse without interpretation? Restaurants without menu schema, hotels without amenity schema, and products without specification schema are harder to recommend correctly.
4. 4.Authority signals — Is the brand mentioned by independent, credible sources in the right context? Agents weight third-party validation heavily.
5. 5.Specificity — Agents default to the brand with the most precise information. "A premium dining experience" loses to "a 40-seat chef-driven restaurant known for wood-fired proteins and a 300-label wine program."

The Window Is Open

The brands investing in AI visibility today are building structural advantages over brands that wait. This is the same pattern that played out with SEO in the early 2000s. Companies that built content and authority early captured durable positions that took competitors years to close.

The difference now is the timeline is compressed. AI adoption is not following the gradual curve of early SEO. ChatGPT crossed 100 million users in two months, the fastest consumer technology adoption on record. Perplexity is processing hundreds of millions of queries per month. Google AI Overviews now appear above traditional results on a meaningful share of consumer queries, and independent CTR studies have shown click-through to organic results dropping when those overviews appear.

The question is not whether your customers will use AI agents to discover and purchase. They already are. The question is whether your brand is the answer when they do.

What to Do About It

The starting point is an honest audit. What do major AI systems say about your brand when asked the queries that matter most to your category? Where do they recommend competitors instead? What information gaps are preventing confident recommendations?

From there the work is methodical. Close the gaps. Structure your content for machine comprehension. Build the authority signals agent systems weight. Monitor your visibility as platforms evolve.

The brands that do this work now will not be playing catch-up in three years. The brands that do not will be explaining to their teams why a competitor with an objectively inferior product keeps getting recommended instead.

Sources

• McKinsey & Company (2024). The economic potential of generative AI: The next productivity frontier.
• Reuters / UBS (2023). ChatGPT user growth analysis.
• Adobe Analytics (2024). Generative AI traffic to retail sites, 2024 holiday season.
• Aggarwal, P., Murahari, V., et al. (2024). GEO: Generative Engine Optimization. Princeton University.