Meta Connect kicks off Wednesday, and while the annual event is still associated with virtual reality and smart glasses, this year’s agenda offers plenty for advertisers and marketers to watch.
The reason is AI.
Across sessions on business agents, multimodal models, AI glasses and content creation, Meta is laying out pieces of an AI ecosystem that could change how consumers discover products, interact with businesses and ultimately transact.
Notably, there are no dedicated advertising sessions on the published agenda, but that doesn’t mean there isn’t an advertising story.
Here are four areas worth watching:
1. Business agents move down the funnel
The most directly relevant session is titled “Build AI agents with Meta Business Agent Platform,” led by Danny Douglass, Meta Forward Deployed Engineering (FDE) Leader, Business Agents.
Meta says the platform operates where people and businesses already exchange more than 1 billion messages per day and allows companies to turn messaging threads into an “always-on, full-funnel growth engine” capable of “capturing leads, closing sales, and supporting customers all in the same place.” Douglass reiterated the same point in a recent LinkedIn post.
The language matters. Meta is positioning business agents as more than customer-service bots. They can become conversion environments.
Meta already controls enormous consumer attention across Facebook, Instagram, WhatsApp and Messenger as well as the advertising systems businesses use to acquire customers. Business agents could shorten the path between an ad impression and a transaction by inserting an AI agent into the journey.
The question to watch: How closely will Meta connect its advertising system to Business Agent?
2. AI that can act
Two sessions focus on Meta’s Muse models and emphasize a transition from AI that understands information to AI that takes action.
Meta says Muse Spark treats what it perceives as something “to act on, not just something to caption.” Another session describes Muse Spark and Muse Code as providing intelligence capable of seeing, deciding and acting, with a harness supplying the reliability and auditability needed for production.
That see-decide-act loop is fundamental to agentic advertising.
Marketing systems today already automate bidding, targeting and creative selection. Models capable of reasoning and acting across workflows potentially push automation further into campaign creation, management and optimization.
Connect should provide a better indication of how Meta thinks about the infrastructure underneath those agents, even if the examples aren’t explicitly advertising-related.
3. AI glasses become a platform
Meta has devoted multiple developer sessions to AI glasses, including building web apps for Meta Ray-Ban Display, integrating mobile apps with glasses and designing experiences for devices with and without displays.
For marketers, the important question isn’t whether Meta puts ads on glasses anytime soon.
It’s whether glasses can become a new consumer interface for discovery.
If consumers increasingly ask an AI assistant about objects, places, products and businesses they encounter in the physical world, the interface between consumers and commercial information changes. Search doesn’t necessarily begin with a typed query or a social feed. It can begin with what someone is looking at.
That creates longer-term questions familiar from the emergence of AI search: Which businesses get recommended? What information informs the answer? How do brands make themselves visible to the agent? And eventually, where does paid discovery fit?
4. Creation and distribution compress
Horizon, Meta’s social virtual reality and mobile ecosystem, offers another clue about Meta’s direction.
Meta says new creation capabilities will help developers go from an idea to a game, then distribute what they create across Facebook, Instagram and Horizon. A separate session focuses specifically on discovery and audience growth across those surfaces.
The advertising parallel is hard to miss. Meta has already been using AI to reduce the work required to create, target and optimize advertising. Horizon applies a similar model to content: lower the cost of creation, then use Meta’s distribution infrastructure to find the audience. The creator increasingly supplies the objective and assets while Meta handles more of what happens in between.
Put those pieces together and Meta is building more than AI features. It is expanding the number of surfaces where it can potentially control the loop between discovery, decisioning, distribution and transaction.
Register for the livestream: Meta Connect, Sept. 23-24 – Meta
More: Mark Zuckerberg’s keynote is at 4 p.m. PT on Wednesday, Sept. 23, followed by the developer program Thursday.