Social Rocket
Meta is shifting its Ads Manager from AI recommendations towards agentic campaign work. The Meta AI Business Assistant is set to plan and prepare campaigns, generate creatives, change audiences and adjust budgets. The key limit remains for now: changes are prepared as a preview or draft and must be approved by the advertiser.
The new features were presented on 6 October 2026 around Advertising Week New York. Together with Ads Creative Studio, the Meta Ads MCP Server and new audience tools, this creates an advertising system in which more and more operational steps can be controlled through natural language.
From assistant to working tool
The Meta AI Business Assistant sits directly within the group's advertising interfaces. What's new is above all that it is meant to retain more context about a business. Meta is testing deeper memory of past campaigns, seasonality and defined goals. In addition, brand guidelines, tone of voice and strategy documents should be importable from other AI tools.
The agentic functions build on this. Marketers will be able to ask the assistant to plan a campaign, create images and copy, change targeting or adjust a budget. Before implementation, Meta shows previews and approval steps.
This is a different level of automation from classic recommendation cards. The AI doesn't just suggest one parameter; it can prepare an entire workflow.
Meta reports five percent more conversions at the same cost per conversion for users of its previous Business Assistant recommendations. This figure comes from Meta's own product data and is not an independent benchmark.
Creative analysis and campaign control are converging
In parallel, Meta is rolling out Ads Creative Studio more widely. The creative insights gained there are meant to flow directly into the Business Assistant. A team could, in theory, have it analyse which motifs work, create new variants from them and then prepare campaign parameters in the same dialogue.
Generative video advertising is also being expanded. Meta is making the conversion of static images into videos within Advantage+ Creative more widely available. In Ads Creative Studio, variants can be created and elements such as hook, audio, clip length, text overlays or end card changed. For product catalogues, Meta is additionally testing scaled product videos.
This builds on the existing expansion of Meta's advertising automation. In our article Meta Ads 2026: AI creatives grow, attribution is reorganised we have already shown how closely creative production and measurement are moving together.
Meta opens Ads Manager to agents via MCP
For larger advertisers and agencies, the Meta Ads MCP Server is particularly relevant. It connects external AI tools with advertising workflows. According to Meta, agents can there, among other things, upload creatives, retrieve ad previews and evaluate performance charts.
Prepared work appears as a draft in Ads Manager. Meta emphasises that actions are only possible with the permissions you have set up, and they require explicit approval before going live.
This separation is what matters. Agentic advertising does not mean a language model should spend money uncontrolled. The more campaign work is automated, the more important roles, approval rights and traceable boundaries become.
How closely the big platforms are now moving together is shown by the comparison with Amazon Ads Agent and TikTok's Agentic Hub. The interface differs, but the direction is the same: less manual clicking, more goal-setting for an automated system.
Targeting is also becoming more dialogue-driven
Meta is also testing AI Audience Discovery for Detailed Targeting. Advertisers should be able to describe their desired audience in natural language. Ads Manager then suggests matching audience options, from which a person selects.
According to Meta, the feature is due to become available in Detailed Targeting by the end of 2026. In parallel, the Customer Lifecycle Strategy is open to all advertisers. Campaigns can thereby focus more on new customers by making it easier to exclude existing ones.
This fits the shift away from ever more granular interest targeting. We explain in detail why broad audiences, conversion signals and Advantage+ are already becoming more important in Facebook Ads Targeting 2026.
For German teams, governance becomes part of the campaign
The practical question is therefore no longer just whether Meta uses AI. What matters is which tasks a team actually delegates.
Four controls are becoming more important:
- Budget limits: Automated changes should have fixed limits and escalation rules.
- Approvals: Creatives, audiences and larger budget changes need clearly defined owners.
- Brand rules: Imported brand guidelines only help if they are current and unambiguous.
- Data access: First-party data, CRM signals and external AI tools must fit your own data protection and permission concepts.
In Europe in particular, this includes documenting AI and data processes. Our social media compliance checklist 2026 bundles the most important operational controls.
With the new features, Meta takes another step from automated campaign types towards an AI-assisted workflow. Ads Manager does not disappear. It increasingly becomes the place where agents prepare work and people review and approve it.