Bots got most of the attention this month, and a real shift is underway in how we work with AI: we used to chat with it (I mean, I still chat with it like it’s my only friend), and now we schedule it to help with some work. Here are the five most impactful stories for online advertisers for the month of October.
1. OpenAI tests visual ChatGPT ads during image generation
Source: OpenAI | October 5, 2026
OpenAI announced a new visual ad format that it will test during image generation in ChatGPT for users on the Free and Go plans, starting later in October in the US with an initial group of advertisers. The ads are clearly labeled and stay separate from the image the user is creating. OpenAI also added conversion integrations with Hightouch, Tealium, and LiveRamp, more attribution partners, and Negative Phrases for qualifying advertisers. According to DV Rockerbox, WeightWatchers’ attributed cost per acquisition on ChatGPT Ads came in 15.3% lower than its blended paid-search benchmark.
| JumpFly Takeaway |
| I haven’t seen one of these ads yet, but I do like the concept. I get a weekly ‘What’s New in ChatGPT Ads’ email; some things are minor, but this one caught my attention as it is a totally different ad type. I do wonder how aligned the ads will be with the users’ “history/memory” or if they will be totally random. OpenAI describes this format as images of product inspiration, product usage, and the experiences a product makes possible…whatever all that means. I really hope the ads are relevant. I would test different types of creative from lifestyle to product shots and even text-based imagery. We don’t know what will work best on this platform, so you might want to begin planning your approach. Once I see this available in any accounts I manage, I will definitely give this a try to see how it performs vs. the traditional ads. |
2. Always-on AI agents arrive for performance marketers
Source: xAI | October 5, 2026; OpenAI | September 29, 2026
xAI published a guide from the person who runs paid media for Grok Bot, showing how the team set up a new ad program with bots handling demand research, budget and media mix drafts, campaign structure, conversion tracking checks, and reporting. Once campaigns are live, a bot checks spend and conversions by channel every morning and pings the team only when something is off, such as a channel spending double its daily budget overnight or conversions dropping to zero because a tag stopped firing. The guide says to start with reporting and to keep platform actions as drafts for the first few weeks. Six days earlier, OpenAI introduced dots, always-on agents in ChatGPT powered by GPT-6 Astra that have their own cloud computer, connect to over 4,000 apps through plugins, and are rolling out to Pro and Business Premium users in eligible markets.
| JumpFly Takeaway |
| Bots, bots, bots. It sure felt like a new bot was announced almost every week for the past month. As I write this, Anthropic just announced Claude Duet. I signed up for the beta and am waiting to get in. All this bot speak and bot launches point to a shift in how we use LLMs: we started by chatting and asking, and now we can schedule the work, let the bot do it, and hear back only when something important pops up. Agents can take action in ad accounts now, but I’m staying away from that. I don’t think I trust a bot to move money around correctly or to know the ins/outs of a business and other variables it just cannot understand like a human can (well, not yet). My core use case is a competitive intel watch in Grok Bot. It monitors one of the accounts I manage as well as its competitors. The goal is to understand how competitors position themselves in their ad copy, track how their messaging and promotional offers change over time, and compare their strategies and offers to my own. Grok Bot runs a live Google search and takes screenshots of what it sees, so nothing is made up. This was a request I gave it. I chose public information on purpose, and it’s an easy, low-friction way to start with bots that go out and do work for you. |
3. Google tests ads in AI Mode for users with Personal Intelligence turned on
Source: Search Engine Land | October 16, 2026
Google confirmed it is testing ads in AI Mode for users who connect their apps through Personal Intelligence, reversing the ad-free approach it launched with. Under the test, the insights and summaries from connected Google apps that already shape AI Mode responses may also be used to personalize the ads those users see. Google’s own Search Help page spells it out: connected-app “summaries, excerpts and inferences” from those sources “may be used to show you ads,” and “Ads will not be based directly on your Gmail inbox or your Google Photos library.” Google says advertisers cannot access connected apps, sensitive information will not be used to personalize ads, and users must opt in and can disconnect at any time. Google has not announced new advertiser controls, targeting options, or a wider rollout for the test.
| JumpFly Takeaway |
| I use AI Mode all the time, and I am that 1% that trusts every answer because I haven’t been lied to or caught an error… yet. I know others have reported poor quality or even incorrect answers, but that just has not been my experience. I have not yet seen an ad in AI Mode, or maybe I just haven’t been paying attention. A personal ad beats a random ad, and it always intrigues me when some of my friends complain about seeing ads in the first place, because ads are what keep these tools we use daily free. Rant over. I’m unsure how I feel about ads using context from Gmail and Photos, and I need to see my personal intelligence in the wild before I can form a real opinion on this setting. I will say I do prefer personalized over generic, random ads. |
4. Amazon Ads Agent merges the DSP and Ads Console, and Full-Funnel Campaigns post beta results
Source: Amazon Ads | September 30, 2026
At unBoxed 2026, Amazon introduced Amazon Ads Agent, a unified buying experience that brings together what was formerly the DSP and Ads Console, with sponsored ads, Full-Funnel Campaigns, and DVA+ in one platform. A conversational AI experience inside Ads Agent, now in expanded closed beta, lets advertisers ask questions, get recommendations, and take action, and Amazon says it puts nearly 1,000 metrics one natural-language question away. With Full-Funnel Campaigns, advertisers set a goal, and AI handles the channel mix, creative, audiences, and optimization across video, display, and audio. Amazon says beta testers saw 67% higher long-term return on ad spend and 29% lower cost to acquire new-to-brand customers. Those figures are Amazon’s own beta results.
| JumpFly Takeaway |
| I don’t cover a ton of Amazon Ads, and I wanted to bring in the pros on all things marketplace, so I asked Nick Pilcher, our Director of Marketplace, for his take. “I think Amazon is moving in the right direction with AI. We’re already using AI quite a bit internally to analyze advertising performance, identify optimization opportunities, and streamline account management. I think the biggest opportunity is making all of Amazon’s data easier to access and actually use. My only hesitation is that AI recommendations don’t always account for things like profitability, inventory, or broader business goals. Overall, I think it’s a great step forward, especially if it allows us to spend less time digging through data and more time focusing on strategy.” As I’ve come to say a ton in this blog series: trust but verify. Before you accept an Ads Agent recommendation, hold it up against margin, stock levels, and your goals. |
5. OpenAI retires custom GPTs on December 11 and offers one-click migration to plugins
Source: OpenAI | September 11, 2026
OpenAI’s September 11 release notes announced plans to “retire custom GPTs across ChatGPT plans and provide a migration path to plugins.” Its migration FAQ sets the standard retirement date for December 11, 2026, with February 11, 2027 for Enterprise workspaces that have an approved deferral, and plans to end new custom GPT creation in Enterprise workspaces. Creators migrate from My GPTs with a Migrate to plugin button, which uses the GPT’s latest published version: instructions become a skill, knowledge files are copied as reference files, and connected apps come along as apps. Custom actions do not transfer and have to be rebuilt, the selected model doesn’t carry over, and the new plugin starts private, so sharing has to be set up again. After migration, the original GPT becomes read-only, and at retirement, custom GPTs and their pages become inaccessible.
| JumpFly Takeaway |
| This is an FYI for anyone who missed the notice. Custom GPTs are common within my team for various projects. The simplest one we use is an ad copy creator, tailored to the way we like ad copy written instead of a generic prompt. The owner will need to migrate it or recreate it as a plugin before the deadline. Only GPT creators (or a workspace admin on Enterprise) can migrate it. The first thing I recommend is to go through your list of GPTs your team relies on and begin migrating or recreating. I’d focus energy on the more used GPTs. Once you migrate or recreate, test it to make sure you continue to get the same quality output. Refine it from there. I feel I actually use plugins more than a GPT due to the fact that you can connect other tools to really enhance the output. I don’t see phasing out Custom GPTs as a bad thing; it is just an evolution of platforms and best practices. |
Looking Ahead
Bots are getting all the attention right now. I, for one, am a fan based on my experience and think it will become common to have bot co-workers (as weird as that sounds). As marketers, we’ll adapt to these new tools the same way we adapt every time a platform or algorithm change occurs.
Questions about how these AI trends affect your campaigns? Let’s talk.
Read our previous article, AI in Online Advertising: 5 Key Trends from September 2026, and stay tuned for next month’s update on AI in Online Advertising.
