When the customer is an AI: what agentic commerce means for programmatic

AI shopping agents aren't a thought experiment anymore. They're already reshaping how people find products and check out. The traditional funnel of search → click → buy is getting compressed, and in some cases skipped entirely. That changes which channels and signals matter. Brand-building, structured data, and attention-heavy environments like CTV, audio, and in-app are getting more valuable. Advertisers who optimize only for the last click risk going invisible to the next generation of buyers.

The assumption that broke

Digital advertising has run on the same basic story for about twenty years: a person sees an ad, clicks it, lands on a page, decides whether to buy. Every metric, optimization model, and budget allocation followed that path.

The story is starting to fall apart.

In late 2025, OpenAI launched Instant Checkout in ChatGPT, starting with US users buying directly from US Etsy sellers and later extending the idea through integrations such as Instacart. Google is also building toward agentic commerce with its Universal Commerce Protocol, developed with partners including Shopify, Etsy, Wayfair, Target and Walmart. McKinsey has projected that AI-driven commerce could influence trillions of dollars in global transaction value before the end of the decade.

None of this means the old journey disappears overnight. But it does mean the click-to-site assumption is no longer the only model worth building around.

When the buyer on the other end of a bid is an agent acting on behalf of a person, almost every assumption of the programmatic industry needs to be re-examined.

What "agentic commerce" actually is

Agentic commerce is online shopping where an AI agent - operating inside a chat interface, browser, or operating system - autonomously discovers products, compares options, and completes purchases on a user's behalf. The user gives intent (i.e: "find the best wireless headphones under $200, here by Friday") and a few constraints, the agent does the rest.

Adoption is already visible, but full autonomy is still early: NIQ found that 42% of consumers used at least one AI tool to shop in the past month, while 10% engaged with an AI-powered shopping assistant and 5% used a fully autonomous AI agent to place an order on their behalf.

Three things make this genuinely different from earlier AI hype cycles:

1. End-to-end execution - The agent doesn’t just recommend, it transacts.
2. No browser session in the traditional sense - Purchases happen through structured APIs, not pages a human would look at.
3. No click in the traditional sense - The agent can read structured data, evaluate offers, and execute without ever opening a product page a human would recognize.

This shift raises the obvious question for anyone buying programmatic: if a paid search result or a display banner isn't on the agent's path to purchase, what role does paid media play?

The answer is more interesting than the headlines suggest.

Where programmatic actually gets stronger

Reading the agentic commerce news, it's easy to assume programmatic is in trouble. For the channels and tactics that line up with how humans actually consume media. For some programmatic advertising channels, the opposite is true:

CTV and streaming - Agents don't watch Netflix or Disney+, people do. As more high-attention inventory moves to ad-supported streaming, CTV becomes one of the most dependable ways to reach a human before an agent ever takes over the buying step.

Mobile in-app - Apps are still a human-first environment. Games, social, news, audio. That's where sustained attention lives, and that's where brand affinity gets built long before a user prompts an assistant.

Programmatic audio and DOOH (digital out-of-home advertising) - These formats never relied on a click. They're built for a world where the click stops being the unit of measurement.

Brand awareness throughout the funnel - When the actual purchase is delegated to an AI, the moment that matters most is the one that decides which brands the user even considers. That moment moves earlier in the journey, and it happens through media, not at checkout. The dollars that used to chase the last click belong in the moments that decide whether an agent is even invited to consider your brand.

The new rules of visibility

Discoverability is no longer just about ranking on Google, it’s about being instantly legible to machines evaluating offers in milliseconds.

For an AI agent to engage with your inventory instead of routing around it, structured signals like clean product feeds, accurate IAB taxonomies, and verified sellers.json declarations - are now mandatory rather than just good hygiene.

Beyond technical setup, this shift elevates first-party data to your absolute foundation, giving brands with direct customer relationships a proven ROAS advantage over legacy cookie strategies.

Finally, because autonomous systems will amplify whatever supply they are pointed at, curating transparent, high-quality supply paths is the only way to drive true AI efficiency rather than amplifying programmatic waste.

A 2026 playbook

If your media plan still treats every dollar as a bid for the last click, you're fighting yesterday's battle. A more durable plan looks roughly like this:

1. Rebalance toward influence channels. Move some of the budget that historically chased clicks into CTV, audio, in-app, and high-attention placements. Shape consideration before the agent enters the funnel.
2. Make your data agent-legible. Audit product feeds, structured metadata, supply declarations. If a machine can't parse your inventory or your brand, neither can the agents standing between you and your customer.
3. Invest in transparency. Cleaner supply paths, fewer hops, verifiable schain. The closer your spend lands to real, human-attended impressions, the better it performs, with or without an agent in the loop.
4. Stop measuring like it's 2018. CTR and last-click attribution were already losing predictive power. In an agentic world they'll actively mislead you. Outcome-based and incrementality measurement is the floor now, not a stretch goal.

The future isn't less programmatic. It's smarter programmatic.

The headlines say agents will replace ads. The data says something more specific: agents will replace certain kinds of ads, and they'll raise the value of others. The advertisers winning in 2026 are not the ones racing to feed AI shopping assistants. They are the ones investing in the channels, signals, and supply paths that determine whether their brand is even on the agent's list.

At Mars Media Group, we believe programmatic advertising's next decade will be defined by quality of attention, quality of signal, and quality of supply - not the volume of clicks. We are investing in agentic-ready infrastructure, mobile-first targeting, and curated, transparent supply across CTV, audio, in-app and display.

If you want to make sure your brand is the one the AI is suggesting, not the one it is skipping, get in touch.