Agentic AI is rapidly reshaping advertising, and some of the most important changes are happening in places we don’t usually look first. In a recent conversation with ad tech veteran Karsten Weide, we explored how AI is transforming not just targeting or bidding, but the workflows, economics, and even the structure of commerce itself.
While the industry loves to talk about putting the right ad in front of the right person at the right time, Karsten argued that the biggest upside of AI today is not in marginally better targeting but in massive workflow efficiencies. Teams that once needed 200 people can now operate with 50, thanks to automation in planning, trafficking, optimization, and reporting. By contrast, the performance lift on campaign effectiveness itself is often modest because machine learning has already been embedded in ad platforms since at least 2012. The real competitive edge now comes from how well marketers can orchestrate humans, agents, and data into a coherent system.
This observation sits alongside a persistent tension between relevance and privacy. Consumers say they want more relevant ads but are wary of how much data they share. Karsten shared an anecdote from a radio show in Santa Cruz: a caller condemned ad-supported platforms like Google, but when asked what he’d actually pay each month for an ad-free search experience, he went silent. That silence captures the reality of the value exchange: people may dislike advertising in theory, but in practice they often prefer ads to paying full freight for content and services. You can see this in streaming, where ad-supported tiers have proven extremely popular. In AI-native environments like chatbots and AI search, however, the stakes are higher. Users reveal their most intimate intents and preferences, which could power incredibly effective direct response advertising, but the social and regulatory norms around using that data are still unsettled.
We also discussed the “black boxification” of targeting as platforms like Google and Meta push fully automated campaign types. Advertisers plug money into a system that often outperforms their manual setups, yet they lose visibility into how decisions are being made. Large brands and holding companies can still demand controls and transparency; smaller advertisers, lacking both leverage and expertise, are simply asked to trust the box. For them, AI is both a huge advantage (in terms of unlocking sophisticated creative and optimization they could never build alone) and a potential source of deep dependency and platform lock-in.
Perhaps the most disruptive idea we explored was agentic commerce. Instead of personally spending an hour searching for the best deal on a specific pair of sneakers, a semi-autonomous shopping bot could handle the entire process, scanning dozens of retailers and applying your preferences for price, delivery, and returns. Once bots do the shopping, what happens to today’s dominant direct response formats—search ads, retail media, chatbot “sponsored answers”? If human decision-making at the point of intent is replaced by agents, then influencing the human earlier through brand-building becomes even more important. Your bot’s rules and defaults will reflect your existing brand preferences. That suggests a potential renaissance for brand advertising and the emergence of a new category Karsten calls “botvertising,” where marketers send offers directly to agents while they negotiate on a user’s behalf. The infrastructure for this future doesn’t exist yet, but the direction is clear: agentic AI is poised to transform how we work, how we reach people, and ultimately how buying and selling happens online.