How AI in Digital Advertising Is Reshaping Social Commerce


For most of digital advertising’s history, discovery and purchase lived in different places. People found a product in one channel, researched it in another, and bought it somewhere else, often after a long chain of clicks, forms, and checkout pages. Every step in between was a chance to lose them.
Social platforms are closing that gap. On TikTok, a single moment can hold the whole journey: people discover a product, watch a creator explain it, and buy it without leaving the app. A platform once seen mainly as a place to build awareness has become a place where brands drive real business results.
Two forces are making that possible. Creators give products a voice audiences already trust. And AI in digital advertising is removing friction from nearly every step in between, from producing creative to finding the right audience to measuring what worked.
In this episode of AdTech Heroes, our host Jon Wallett sat down with David Kaufman, VP of Product Management and Head of Monetization Product Solutions & Operations at TikTok, to explore how those forces come together.
They talked about integrated commerce, how brands of every size are using TikTok Shop, and how automation and agentic AI are reshaping creative production, media buying, and measurement. At the center of it all is the goal that shapes Kaufman’s work: removing every bit of friction between a client’s investment and the business result it’s meant to drive.
TikTok’s roots are in short-form video, and early on, brands saw it mainly as a way to build brand equity. Today, the platform supports a much wider range of business objectives, and more of them are about performance.
TikTok Shop has made social commerce part of everyday shopping.
Kaufman walked through the traditional online journey: see an ad, click it, visit a website, create an account, enter payment details, add to cart, check out. Every one of those steps is friction. Every step also loses people along the way, which weighs on engagement and conversion rates. On TikTok Shop, most of those steps disappear. People see a product, like it, tap it, and buy it.
The power of commerce at the point of discovery, he said, can’t be overstated.
It also changes what’s hard about selling online. Earlier in his career, as an entrepreneur, Kaufman worked in distributed e-commerce, sharing product data across the social web. Many of his clients believed building the website was the hardest part. It wasn’t. On TikTok Shop, the platform, its users, and its creator ecosystem do much of that work, helping brands find their audience faster and see results sooner.
“Finding your audience is the hard part.”
Not every brand comes to TikTok Shop for the same reason. Kaufman described three distinct patterns, each shaped by the size and mechanics of the business.
Small and medium-sized businesses often treat it as another sales channel, and sometimes the main one. Around seven and a half million SMBs in the US use TikTok solutions. Some run their own websites, and some don’t. Either way, integrated commerce is easy to set up, and sales happen naturally as people discover products in their feed.
Established retailers tend to enter one category at a time. Health and beauty has been especially strong, which fits how much of the platform’s content already lives there. “Get ready with me” videos and hair and makeup tutorials create a natural home for products people want to see in action.
The third pattern may be the most surprising.
Large CPG (consumer packaged goods) companies without a direct-to-consumer presence are using TikTok Shop as a product testing ground. How do people respond to a new snack flavor? What kind of engagement does a new beverage concept earn? Instead of waiting on traditional research, these brands get real-time insights from real buyers. Fulfillment is the main challenge on the back end, and brands can either manage it themselves or work with TikTok.
The result is a hybrid of sales and market research, and it isn’t limited to large companies. For emerging brands, a low barrier to entry makes it easier to test an idea before committing to a bigger launch.

If TikTok Shop is where the sale happens, creators are often what make it happen.
Kaufman pointed to TikTok’s creator ecosystem as one of its biggest strengths. Through the TikTok One platform, brands can work directly with creators who have already built trusted voices and the audiences brands want to reach.
That authenticity is changing performance advertising too. With Smart+, TikTok’s automated performance solution, a brand can bring a large product catalog and let the system match the right product to the right person in the right format.
The creative is evolving as well. Five or ten years ago, a product ad might have been a static image against a white background. Now, brands can bring in creators to explain and enrich the product, and TikTok Shop sellers can work with affiliate creators who make content specifically around what they sell.
On a platform like TikTok, creative has become the targeting. Rather than defining every audience in advance, brands can let their message find the people who respond to it. Personalizing content becomes less about who someone is and more about what they choose to engage with.
Behind the creators and the commerce sits a lot of AI. Kaufman’s view of it is a calm one: AI in advertising is not scary at all.
The industry has relied on ranking for years. What we used to call machine learning, we now call AI, and AI models have long decided which ad appears in front of which person. Kaufman called it maybe the most established and valuable use case for AI anywhere.
He remembers when that work was manual. As an entrepreneur, he found those kinds of insights through hands-on data analysis. Today, they live inside AI algorithms, built into how ads are delivered.
What’s new is where AI shows up next. Some use cases are small but practical: fixing aspect ratios, correcting color, or adding promotional overlays across a catalog of hundreds of thousands of assets. These are repetitive tasks that would be nearly impossible to do by hand at scale. Others are more ambitious. Through TikTok’s Symphony platform, AI enables brands to generate scripts and video from a prompt, or to create a digital avatar that explains a product.
Traditional AI works in the background. It ranks, predicts, and optimizes within rules a team has already set. Agentic AI goes a step further. It can understand a goal, plan the work, and act on it.
Kaufman sees TikTok’s role as offering both paths.
A marketer can upload a CRM file of their highest-value customers for a retargeting campaign, or turn on automation and let the system find those customers on its own. Teams that want a deterministic setup can program their business logic against the campaign management API (application programming interface). Others are testing the MCP (Model Context Protocol) offering TikTok announced at TikTok World in May. It lets an agent trained on a brand’s business logic connect directly to the ads platform and manage campaigns.
According to Kaufman, the most sophisticated advertisers are doing a little of everything to learn what works. That’s the real promise of agentic AI for media buying: giving teams the freedom to decide where automation helps and where precision matters more.

Creator marketing has long been one of the most manual parts of social media advertising. Finding the right partner meant researching profiles, scrolling through social media posts, and reaching out directly or tracking down a business manager.
This is where agentic AI in marketing starts to change the work. With natural language processing, a marketer can describe a goal in plain words and get back a shortlist of relevant creators, suggested outreach, and draft briefs. On TikTok, that happens through the Symphony agent, part of what Kaufman calls an agentic buying flow.
What doesn’t change is the need for judgment. An agent can surface options, but teams still decide which creators genuinely fit the brand and which messages will feel authentic to their audience.
AI avatars have a role too, but a different one. If a brand has already defined its audience and wants to test many versions of a message quickly, generative AI makes that possible without a camera crew or a long production timeline. When it comes to authenticity, though, most of the excitement Kaufman hears from advertisers is still about creators.
As commerce and automation grow, so does the need for proof. Sophisticated brands build marketing strategies across every channel, so measurement can’t stop at one platform’s walls.
Measurement was one of the defining topics at Cannes Lions this year, and the message from brands was consistent: they want partners that aren’t checking their own homework. That’s why third-party validation and multi-touch attribution matter as much as the numbers a platform reports on its own.
Kaufman added a principle worth remembering: the client owns the definition of value. For some brands, that means brand lift surveys. For others, it’s media mix models or in-house business intelligence teams. Either way, the platform’s job is to understand how each advertiser defines success and deliver against it.
Not every outcome is a sale. Some brands need to own a moment.
TopView, the ad people see when they first open the app, is built for advertisers who want a high share of voice around a moment that matters to their business. Pulse Premier lets brands appear alongside content tied to major events such as the World Cup, often to extend sponsorships they run on other channels.
Timing matters too. During the Super Bowl, TikTok found clear spikes in engagement during game breaks, a strong signal for sponsors to run complementary content on the platform. The exception was Bad Bunny’s halftime show, when engagement dipped. Some moments simply command full attention.
It’s an idea we come back to often. Attention clusters around moments, and the brands that understand those moments as they happen are the ones that earn real engagement.
The future trends Kaufman described point in one direction: less friction between discovery and purchase. Creators give products a trusted voice. Integrated commerce turns interest into action in a few taps. And AI adoption is speeding up everything around it, from creative production to buying and measurement.
The brands that grow fastest will treat discovery, creative, and commerce as one connected experience, built around the people they want to reach.
Find answers to get the most out of Seedtag.
Neuro-Contextual Advertising is an evolution of contextual advertising that uses AI to understand interest, emotion, and intent within content. Instead of relying on personal data, it helps brands align advertising with the moment people are experiencing in real time.
Contextual advertising places ads based on the content a person is engaging with, rather than personal browsing history or identifiers. Neuro-Contextual approaches go beyond traditional contextual targeting by understanding audience interest, emotion, and intent.
CTV advertising refers to ads delivered through Connected TV environments, including streaming platforms and smart TV applications. It allows brands to reach audiences across premium video experiences using more contextual and privacy-first targeting approaches.
Liz is Seedtag’s proprietary Neuro-Contextual AI, designed for full-funnel advertising across premium open web, video, and CTV environments. Grounded in neuroscience, Liz understands contextual signals such as interest, emotion, and intent in a human-like way to deliver relevant advertising experiences in real time while respecting user privacy.
With consumers increasingly wary of data collection and cookies, privacy-first advertising focuses on delivering personalized and relevant ads while respecting user privacy. Neuro-Contextual advertising enables relevant advertising experiences without relying on personal data or third-party cookies.
Leveraging neuroscience principles, Liz recognizes patterns, interprets context, and responds dynamically to user interests, emotions, and intentions. This creates a more cohesive and intelligent system capable of delivering more relevant advertising experiences.