TL;DR
The biggest agentic AI marketing news of the past 12 months is that AI agents moved from demos into the places customers shop and advertisers buy. ChatGPT, Google and Meta all built ways for an AI to find or buy products, and Google and Meta put agents inside their ad tools.
But the first in-chat checkout stalled — Walmart said those sales converted at a third of the rate of shoppers sent to its own site — and Gartner finds CMOs say AI automates just 16% of marketing work today. Agents grew fast; the human still makes the call.
If you read marketing news over the past year, you heard “agentic” every week. Most of it was announcements. Some of it shipped, some of it was quietly walked back, and a lot of it is still “in testing.” For a small business owner, it is hard to tell which parts change what you should do on Monday.
This piece sorts the agentic AI marketing news from October 2025 to October 2026 into what launched, what stalled and what the numbers say. Every fact links to its source, and where a figure comes from a company that sells the thing it measured, that is said. If you want the basics first — what an AI agent is and how it differs from a chatbot — start with our plain-English guide to what agentic AI is.
Agentic AI marketing news: what changed in the last 12 months?
Three things changed: AI assistants became places to shop, the big ad platforms built agents into their buying tools, and marketing software rebranded around agents. A person still approves the work. The timeline lists the moments that mattered.
| When | What happened | Why it matters to marketers |
|---|---|---|
| Sept 2025 | OpenAI launches Instant Checkout in ChatGPT, built on the open Agentic Commerce Protocol with Stripe | The first mainstream “buy inside the AI” button |
| Sept–Oct 2025 | HubSpot shows a wave of Breeze agents at INBOUND; Salesforce launches Agentforce 360 at Dreamforce | Marketing software starts selling “agents”, mostly in beta |
| Dec 2025 | Salesforce says AI and agents influenced about 20% of global Cyber Week orders | A vendor figure, but the first big holiday measured this way |
| Jan 2026 | Adobe reports AI-referred retail traffic up 693% over the holidays; Google announces the Universal Commerce Protocol | Shoppers are arriving from AI tools; Google sets its own standard |
| Feb 2026 | Shoppers can buy from Etsy and Wayfair inside Google’s AI Mode; ChatGPT starts testing ads in the US | AI answers become both a shop and an ad channel |
| Mar 2026 | OpenAI moves Instant Checkout into retailer apps; Walmart says in-chat sales converted at a third of the rate | The first checkout model stalls |
| May 2026 | Google Marketing Live unveils Ask Advisor, an agent across Google Ads, Analytics and Merchant Center | An agent inside the ad account itself |
| Oct 2026 | Meta begins testing a business assistant that can plan campaigns and change targeting and budgets | Agents start touching the money, with approval |
AI shopping agents arrived — and the first checkout stumbled
Agentic commerce means an AI assistant finding, comparing and sometimes buying a product for a shopper. On 29 September 2025, OpenAI launched Instant Checkout, letting US ChatGPT users buy single items from Etsy sellers inside the chat, with Shopify merchants promised next. It ran on the Agentic Commerce Protocol, an open standard OpenAI built with Stripe so that an AI agent and a shop’s back end can talk to each other.
Google answered in January 2026 with the Universal Commerce Protocol, developed with Shopify, Etsy, Wayfair, Target and Walmart. By February, US shoppers could buy from Etsy and Wayfair inside Google’s AI Mode.
Why in-chat checkout stalled
In March 2026, OpenAI told reporters that “Instant Checkout is moving to Apps” — shoppers would finish buying in each retailer’s own app inside ChatGPT. The reason came from Walmart, which had offered around 200,000 products through the feature: its executive Daniel Danker said purchases made inside the chat converted at one-third the rate of shoppers who clicked through to Walmart’s site, as reported by Search Engine Land. The open protocol survived; the idea of the AI owning the whole sale did not, at least not yet.
But AI-referred shoppers kept growing
The checkout stalled; the discovery did not. Adobe Analytics found that traffic to US retail sites from generative AI tools rose 693.4% between 1 November and 31 December 2025 compared with a year earlier — while noting that the base of users “remains modest.” Salesforce estimated that AI and agents influenced 20% of global Cyber Week orders, about $67 billion, through recommendations and service chats. Salesforce sells agents and modelled that figure, so treat it as an indication.
How are AI agents being used in advertising now?
AI agents in advertising now sit inside the ad accounts themselves, suggesting and — with permission — making changes. A year ago, ad automation mostly meant the platform choosing audiences and placements. Over the past 12 months, Google and Meta both added assistants that can be asked questions in plain words and act on the account.
- Google, May 2026 — at Google Marketing Live, Google introduced Ask Advisor, which it calls “a new, unified agent” spanning Google Ads, Google Analytics, Merchant Center and Google Marketing Platform, alongside new ad formats for AI search and more of its automated AI Max campaigns.
- Meta, October 2026 — at Advertising Week New York, Meta said its business assistant is “beginning to be tested” for planning campaigns, changing targeting and adjusting budgets, with advertisers able to approve or limit those actions. Its expanded Ads MCP server lets outside AI tools stage work that lands as a draft in Ads Manager (Social Media Today).
- OpenAI, February 2026 onward — ChatGPT began testing ads for logged-in US adults on its Free and Go plans, shown below the answer, and has since expanded the test to more countries. Not an agent, but a sign that AI answers are becoming a paid channel.
Meta says advertisers using its assistant are “typically seeing 5% more conversions at the same cost per conversion,” but has not published how it measured that. Meta was reported in June 2025 to be aiming for fully automated ad campaigns by the end of 2026; as of October 2026, the parts that change budgets and targeting are still in testing.
Agentic AI statistics: how many marketers actually use agents?
Fewer than the headlines suggest. In a Gartner survey of 402 CMOs fielded from August to October 2025 and published in May 2026, marketing leaders said AI automates 16% of marketing work in 2026 and expect that to reach 36% by 2028. That is real growth, and it still leaves most of the work with people.
| Source | What it measured | Finding |
|---|---|---|
| Gartner, May 2026 | 402 CMOs, surveyed Aug–Oct 2025 | AI automates 16% of marketing work in 2026; 36% expected by 2028 |
| McKinsey, Nov 2025 | Global State of AI survey | 62% of organizations at least experimenting with AI agents; 23% scaling one somewhere; no more than 10% scaling them in any single function |
| Gartner, Jan 2026 | Prediction | 60% of brands will use agentic AI for one-to-one customer interactions by 2028 |
| Gartner, June 2025 | Prediction | Over 40% of agentic AI projects will be canceled by the end of 2027 |
Read together: almost everyone is trying agents, a minority runs them for real, and in any one department — marketing included — the share is small. The cancellation forecast is the useful warning. Gartner’s analysts pointed to rising costs, unclear business value and weak risk controls, and to “agent washing” — older products relabelled as agents. Our earlier piece on the rise of agentic AI in marketing covers the bigger picture behind these numbers.
Marketing software rebuilt itself around agents
The third shift was in the tools marketers already pay for. At INBOUND in September 2025, HubSpot announced a batch of Breeze agents for marketing, sales and service, most of them launched in beta. In October 2025, Salesforce launched Agentforce 360 at Dreamforce, folding its agents, data and apps into one system.
For a small business, “agent” now appears on many pricing pages, and the word alone tells you little. Ask whether it plans the steps itself, uses real tools and data, and where a person signs off — the autonomy ladder in our guide sets this out.
"Shoppers are happy to let AI find the product, and still want to buy it on the shop’s own site."
What didn’t change
- A person still presses go — every ad-platform agent above either is in testing or asks for approval before it spends money.
- Agents are only as good as what they know about you — a thin website or a half-filled listing gets the average guess.
- The shop’s own site still closes the sale — the Walmart figure is one retailer’s, but the clearest public number of the year.
- Most marketing work is still done by people — Gartner’s 16% is the CMOs’ own estimate, and it is the share of work, not of jobs.
What this means for a small business
You don’t need to buy an “agent” to respond to the past year. You need to be easy for agents to read, and clear about what you let them do. Five things worth doing now:
- 01.Make your product data complete — AI shopping agents compare products from the facts they can read: titles, specs, prices, stock and structured product markup on your pages. A listing with gaps is the one an agent skips.
- 02.Write your brand facts down once — who you sell to, what you can prove, the words you never use. Every agent, ours or a platform’s, works better from confirmed facts than from guesses.
- 03.Check AI referrals in your analytics — look for visits from ChatGPT, Perplexity and other AI assistants. If they are growing, treat the pages they land on like ad landing pages.
- 04.Set limits before you switch on an ad agent — decide the budget ceiling and the changes you approve by hand. Both Google and Meta let you keep approval; keep it until you trust the results.
- 05.Judge tools by what they hand you — a finished, checkable piece of work beats a promise of autonomy. Our framework on what to automate in marketing helps decide which steps to hand over.
For the wider plan these steps sit in, our guide to how to use AI for marketing sets out the order of work.
Where TurboAgents fits
TurboAgents makes the marketing work; it doesn’t run your ad account or your shop. That puts it on the “agents hand you finished work, a person decides” side of the year’s news.
- Brand Brain reads your website once and drafts your brand profile, products, customer types and competitors. You confirm what is right, and nearly thirty tools work from those facts.
- The Workflow Builder chains tools into one run, handing each step the full work of the steps before it, with the bill shown before anything is charged.
- The Product Description & SEO Pack writes listings for Amazon, Shopify, Flipkart, Walmart, Etsy and Myntra to each marketplace’s rules, plus ready-to-paste Product schema markup for your own page — the structured facts shopping agents read.
- The Google Ads Architect builds a complete Search campaign and exports it as a Google Ads Editor file you review and import yourself.
Who this doesn’t suit: anyone who wants AI to run their ad spend unattended (TurboAgents hands over a launch-ready campaign; a person still presses go), and teams that need shared seats and approval chains — every plan is single-seat. Plans start at $29 a month for Basic.
Methodology & sources
This review covers announcements and data from late September 2025 to 13 October 2026, all checked on 13 October 2026. Figures are reported as each company stated them; several sources sell the AI products they measure, and so does TurboAgents, which is why every claim links to its source.
- OpenAI — Instant Checkout announcement, 29 September 2025; “Testing ads in ChatGPT”, January 2026, with later updates.
- Walmart via Search Engine Land — in-ChatGPT checkout conversion, March 2026.
- Google — Universal Commerce Protocol coverage (InfoQ), January 2026; Google Marketing Live 2026 announcements, 20 May 2026.
- Meta — Advertising Week New York announcements, 6 October 2026, via Social Media Today and Relevant Audience.
- Adobe Analytics — 2025 holiday shopping season, 7 January 2026: US retail sites, 1 November–31 December 2025.
- Salesforce — Cyber Week data, 5 December 2025: 25 November–1 December 2025, modelled from Salesforce and third-party data.
- Gartner — CMO survey, 11 May 2026; one-to-one interactions prediction, 15 January 2026; agentic AI cancellations prediction, 25 June 2025.
- McKinsey — “The state of AI in 2025: Agents, innovation, and transformation”, November 2025.
- HubSpot — INBOUND 2025 announcement, 3 September 2025.
If the year’s news leaves you with one job, make it this: get your brand and product facts into a shape any agent can read. Set up Brand Brain once, then run the Product Description & SEO Pack on the products you sell most.
Start a free 2-day trial — no credit card required — and see what agents can hand you when they start from your real facts.
Frequently Asked Questions
Agentic commerce is shopping where an AI assistant finds, compares and sometimes buys a product on a person’s behalf. Over the past year OpenAI, Google and Meta each built ways to do this, using open standards such as the Agentic Commerce Protocol and Google’s Universal Commerce Protocol so AI agents can talk to shops’ systems.
Not in its original form. OpenAI did not announce a shutdown, but in March 2026 it said Instant Checkout was moving to apps, so purchases happen inside each retailer’s own app in ChatGPT. Walmart said in-chat purchases had converted at one-third the rate of shoppers sent to its website. The underlying Agentic Commerce Protocol remains an open standard.
Partly, and with your approval. Google introduced Ask Advisor, an agent across Google Ads, Analytics and Merchant Center, in May 2026. In October 2026 Meta began testing a business assistant that can plan campaigns and change targeting and budgets, with advertisers able to approve or limit its actions. Keep a budget ceiling and approve changes yourself.
In a Gartner survey of 402 CMOs fielded in late 2025 and published in May 2026, marketing leaders said AI automates 16% of marketing work in 2026 and expected that to rise to 36% by 2028. That is the CMOs’ own estimate of the share of work, not a count of jobs replaced.
Yes, though from a small base. Adobe Analytics found traffic to US retail sites from generative AI tools rose 693% in November and December 2025 compared with a year earlier, while noting the number of users is still modest. Checking AI referrals in your own analytics shows whether it matters for your shop.
Make your product information complete and structured: clear titles, specifications, prices, stock and product schema markup on your pages, so an agent can read and compare them. Write down your brand facts and proof once, and keep the final purchase on your own site working well, since shoppers still tend to buy there.
