Social Commerce Is Going Agentic: Seller Agents Now, Buyer Agents Next

Social commerce is going agentic, and the money behind it is real. US social commerce passes $100 billion in 2026, up about 18% on the year, with TikTok Shop alone projected near $23 billion in US sales (eMarketer). But the agents arriving in that market are not what most headlines suggest. The first ones work on the seller's side, running the store. The agent that shops for the buyer is still being built.

Getting that order straight tells you what to use now and what to prepare for.

What TikTok's seller agents actually do

TikTok moved first and loudest. On June 30, 2026 it launched the Agentic Hub, a marketplace of AI Skills for advertisers and merchants, built on TikTok's MCP with native and third-party agents (HubSpot, Wix and others among the early partners). The skills cover the real work of running a shop: campaign creation, creative generation, audience analysis, catalog management, performance reporting, and optimization.

Separately, TikTok Shop is rolling out an AI Homepage in the Seller Center, replacing the old module dashboard with five agents that surface shop data, flag issues, and suggest next actions in real time, in beta for a limited set of US sellers. Both are the merchant's helpers. Neither buys on a customer's behalf.

Seller agents already run your social ads

TikTok is not first at this, only loudest this quarter. The biggest seller agents in social have been running quietly inside the ad platforms for a while. Meta's Advantage+ and Google's AI Max and Performance Max already handle targeting, bidding, creative and budget pacing with steadily less human input, and standalone agents like Smartly and Albert run accounts across channels.

The scale is not small. US AI ad spending reaches about $32 billion in 2026, nearly triple the prior year, on its way past $68 billion by 2030 (eMarketer).

“

“We estimate that the US livestreaming e‑commerce market will grow to $68 billion by 2026.”

Meta's Advantage+ suite alone runs at roughly a $60 billion annual pace and, by Meta's own numbers, returns about 22% more than manually built campaigns.

“

“US AI ad spending will reach $32.03 billion in 2026 — nearly triple last year’s total — and will exceed $68 billion by 2030, according to eMarketer’s May forecast.”

If you run paid social, a seller agent is probably already spending your budget.

How much of this is real, and how fast

Take the momentum seriously and keep the hype in check. Agentic AI is still early in most companies: McKinsey finds 62% of organizations experimenting with it and only 23% scaling. The trajectory is the story. The IAB reports two in three ad buyers already testing agentic AI for video, and PubMatic forecasts a quarter of all ad buys running autonomously by 2028 and more than half by 2030. Early, but moving fast enough that waiting is a decision.

What seller agents change for your job

When an agent runs the ads and the shop ops, your job changes shape. You stop doing the tasks and start feeding the machine and supervising it. What the agent needs from you is what the rest of agentic commerce needs: a clean product feed, fresh creative, and accurate conversion data. Garbage in, confident garbage out, at machine scale.

Two cautions. The platforms fold this automation in so deep that opting out gets harder each release, and the agent optimizes for the platform's goal, which is spend and on-platform GMV, not always your margin. Teams doing this well run governed autonomy: spend caps, approval gates, and audit trails, not blind trust. And you have to measure what the agents actually drive, which is the work in the agent-revenue piece, on top of the feed hygiene in the feed requirements piece.

Buyer agents are next: Meta's reported Instagram shopping agent

The buyer side is coming from Meta, but this is reporting, not a release. According to eMarketer and others, Meta is building a consumer agentic assistant, internally codenamed Hatch, plus an Instagram shopping agent that would let people browse, compare, and buy inside the app, reportedly targeted for around Q4 2026 to rival TikTok Shop. It is not live, and the internal specifics are unconfirmed reporting, so treat the whole thing as building, not shipped.

The direction is the signal, not the date: the platforms want an agent that buys for the consumer, on-platform.

How to get ready for social buyer agents before they ship

1. Use the seller agents that are here. TikTok's Agentic Hub and AI Homepage, plus Meta and Google's ad automation, are live today. Adopt them for the efficiency, with eyes open that they serve the merchant, and cap what they can decide alone.
2. Get your product data agent-ready. When buyer agents ship, they will buy the way every agent buys, from clean, structured product data. That is the work in the feed requirements piece.
3. Do not go social-exclusive. A product that lives only inside a social app is one a future buyer agent may not reach from your crawlable surface, the citation trap from the social visibility piece.
4. Prepare data, not one integration. Meta will not be the only buyer agent, so build the feed that works for all of them rather than wiring your store to a single platform's agent.

Social commerce going agentic is really two waves: seller agents you can use today, buyer agents you should be ready for. The prep is the same clean, reachable product data the rest of agentic commerce runs on, and the full picture is in the pillar guide.

Want to check whether an AI agent can already find and buy your products? Run GEOlikeaPro's Visibility Vitals checker and start with the feed.