AI is reshaping ad creative and product discovery faster than most marketing organizations are prepared for

Speechify is already testing close to 1,000 AI-generated ads per day. Cliff Weitzman, the company’s founder, puts the number alongside a separate figure: roughly 8,000 human-made creatives produced monthly. That combination, AI volume layered on top of a substantial human baseline, is not a pilot. It is a production operation, and it signals how quickly the creative economics of digital advertising are shifting.

The structural argument for why this matters at scale comes from Rory O’Driscoll, a venture investor. O’Driscoll’s framing is Darwinian: if there is a correlation between AI-driven campaign capability and commercial success, then competitive pressure will do the sorting. “That 10 will become 20 will become 40,” he says, “because the people who can’t do it will be forced out.” The point is not that every marketing organization will adopt AI willingly. It is that the ones that do not will lose ground to the ones that do.

The timeline for the next wave of capability comes from Liva Ralaivola, a researcher working on AI systems for advertising. Ralaivola projects that real-time generative ad creative will be feasible within roughly two to three years. The further horizon he describes is more structurally significant: within five years, users may find themselves telling AI assistants to surface advertising on request, actively consenting to curated product options as part of a conversational interaction. That is a different relationship between consumer and ad than the industry has operated on for decades.

That 10 will become 20 will become 40, because the people who can't do it will be forced out. Rory O'Driscoll

The product-discovery side of this shift is already producing measurable results. Andy Fang, a co-founder of DoorDash, reports that 50 percent of Ask DoorDash restaurant order trajectories result in users ordering from restaurants they have never ordered from before. On the grocery side, AI-assisted orders are running at basket sizes 40 percent larger. These are not engagement metrics. They are purchase outcomes, and they suggest that AI curation is already changing what people buy, not just what they see.

Martin Casado, a general partner at Andreessen Horowitz, adds the image-based query as another dimension of capability now operating in the market. He describes the use case precisely: a user photographs a coat and asks what it is and where to buy it. Five years ago, Casado says, that probably would not have worked. It works now. The gap between what consumers can do with a camera and a search interface and what they could do a few years ago has closed in a way that opens new paths between intent and purchase.

What the evidence describes is a compounding shift across three layers simultaneously: creative production, where volume and variation now cost a fraction of what human-only workflows require; discovery, where AI curation is demonstrably changing which products consumers find and buy; and consent, where the longer-term model points toward users requesting advertising rather than tolerating it. Each of those shifts is meaningful on its own. Together, they suggest that the advertising industry is not simply adding AI tools to existing processes. It is moving toward a different underlying architecture, one where the campaign as a fixed, pre-produced artifact gives way to something closer to continuous, personalized generation. The organizations best positioned for that transition are the ones that treat current experimentation as structural preparation, not optimization.