Meta’s business assistant inside Ads Manager now keeps a memory of what an advertiser tried last quarter, its seasonality and its stated goals. The features that would let it change a campaign from a chat prompt are still in testing.
Meta announced the update at Advertising Week New York on Oct. 6. Its English announcement was not available to us when we wrote this, so we worked from the post on Meta’s Brazil newsroom, which is in Portuguese. Translations of its wording are ours.
What shipped and what is still in testing
Meta’s post separates the announcements by stage, and the stage matters more than the headline. These are available now or rolling out:
- Ads Creative Studio, the creative workspace Meta introduced a few months ago, has begun rolling out to a larger group of advertisers after initial testing.
- The Meta AI business assistant, which Meta abbreviates MAIBA, now has deeper memory of a business, including the previous quarter’s history, brand seasonality and defined goals. It also takes creative insights from Ads Creative Studio.
- Customer Lifecycle Strategy is available to all advertisers.
- In Advantage+ creative, advertisers and agencies can generate video from existing static images, a feature Meta describes as generally available.
These are in testing or scheduled:
- Importing brand guidelines, tone of voice and strategies from other AI tools a business already uses.
- Agentic functions in the assistant. Meta says that soon it will be possible to plan and create a campaign, generate images and copy, update audience targeting or change budgets conversationally.
- Scheduled tasks and automated alerts for routines such as performance monitoring, weekly summaries and budget-review reminders.
- AI audience discovery in Detailed Targeting, which Meta says arrives by the end of the year.
- An Audience Label that would exclude a whole group of existing customers with one tag.
The agent layer: approvals and the MCP server
Meta says previews, approvals and controls let an advertiser decide what to delegate to the AI and what to keep under human control. For large advertisers and agencies, the Meta Ads MCP server now supports more workflows inside the AI tools they already use. Marketers can upload creative, preview how ads will look on Facebook and Instagram, and follow performance through trend charts and breakdowns.
The detail that deserves attention is where agent work lands. Per the post, work prepared by agents appears as a draft in Ads Manager, so people and AI can co-author the same campaign and publish from either interface. Meta adds that all actions depend on permissions the advertiser configures and on explicit approval before going live. This is the same control pattern we noted when Meta connected its Muse agent to Klaviyo and Shopify, where nothing publishes, sends or spends without the user’s approval.
Audiences and video
AI audience discovery lets a brand or agency describe its ideal customer in natural language rather than scroll through targeting options. Ads Manager then surfaces the most relevant audiences in seconds, and the advertiser keeps only the ones it wants.
Customer Lifecycle Strategy addresses a narrower job. When an advertiser marks a campaign as acquisition, the tool recommends excluding existing customers, so the team does not build those exclusion lists from scratch.
On video, Meta cites its own figure that more than 60% of the time people spend on Facebook and Instagram goes to watching video. Beyond the static-to-video feature, Ads Creative Studio can generate new video concepts and variations and edit hooks, audio, clip length, text overlays, captions and end scenes. Advertisers with product catalogs in the beta program can turn images into product videos at scale, and Meta says smarter product recommendations that read the ad image, copy and landing page will expand in the coming months.
The one quote in the post
Meta’s post includes a customer endorsement from Whitney Mortimer, VP and Executive Creative Director at Tinuiti: “O Ads Creative Studio da Meta é uma base sólida para a otimização criativa.” In English, she calls the tool a solid base for creative optimization, and says it lets her team identify the best-performing creative by campaign objective.
What it means for the marketing leader
The sections below are our reading, separate from Meta’s announcement.
Start with memory. An assistant that remembers last quarter and your stated goals will give advice shaped by whatever those records say. If the goals on file are stale, the recommendations will carry that forward. Someone on the team should own what the assistant knows about the account and review it each quarter.
Next, settle delegation tiers before the agentic features reach your account. Meta’s own design assumes the advertiser decides what to hand over. A workable split puts reporting and alerts first, creative drafts second, and budget and targeting changes last, with a named approver for each. The governance gap we described in GTM Teams Deploy AI Agents Faster Than Governance opens when tools arrive before those decisions are made.
Third, treat Meta’s performance claim as a number to test. The post says advertisers who adopt the assistant’s recommendations typically see 5% more conversions at the same cost per conversion than those who do not. That is Meta’s statement about its own customers. A holdout in your own account, with some campaigns following recommendations and some not, will tell you what it is worth to you.
Finally, note where the MCP server sits. Agencies and large advertisers can now run Meta campaign work from third-party AI tools, with drafts landing in Ads Manager. That makes Ads Manager the approval point even when the work starts elsewhere. We covered how Meta is also selling its AI stack to businesses in Meta Opens an Enterprise Platform for Its AI Stack, and the buying-side shift in Amazon, Meta, Discord and Google Hand More Ad Buying to Software.
Before the next quarterly plan, list which of the tested features you would switch on, who approves each, and what result would make you turn it off.
Source: Meta
