Meta wants advertising on its platforms to be as simple as entering a credit card number and a business goal. AI handles the rest.
That's the vision Mark Zuckerberg has publicly committed to. And with Meta's Andromeda ad retrieval system now powering delivery across Facebook and Instagram, the company is closer to that future than most advertisers realize.
But the path from here to there is creating a tension that every marketing team running paid social needs to understand — because the tradeoffs aren't just technical. They're strategic.
What Andromeda actually changes
Andromeda is a full rebuild of Meta's ad delivery engine. It replaced the legacy system that matched ads to users based primarily on audience targeting inputs from advertisers. The new system scans millions of ad options per second, using AI to predict the best match between creative and user — regardless of how narrow or broad the advertiser's targeting is.
The results are measurable. Changes made to Andromeda in Q3 2025 produced a 14% improvement in ad quality on Facebook, with an 8% quality improvement on selected segments across Instagram. The system incentivizes uploading more diverse creative and showing it to broader audiences, then letting AI determine which messages and formats resonate with which users.
For advertisers, this fundamentally shifts the lever of control. The old model was: define your audience, create ads for that audience, optimize delivery. The new model is: provide many creative variations, let AI find the audience for each one.
The scale of the shift
The adoption numbers tell the story of how fast this is moving. At one major agency, Advantage+ campaigns — Meta's automation suite powered by Andromeda — now represent 60-70% of their total Meta spending. That's not a test budget. That's the majority of spend on one of the world's largest advertising platforms being managed by AI systems with limited advertiser visibility into targeting and allocation decisions.
The creative demands have scaled accordingly. One client project that previously required 300 creative assets now demands 1,000. Andromeda's model — test more creative variations across broader audiences — means creative production volume has become a strategic input in a way it never was before.
The control problem
Here's where it gets complicated. Agencies working with Meta's AI tools describe the experience as playing "Whac-A-Mole" — new automation features appear frequently, often with limited transparency about what they change and how they affect campaign performance.
There is meaningful hesitation among marketers about fully embracing AI-generated creative. Clients resist the pitch that AI creative enhances rather than replaces brand work. The concern isn't irrational — when the platform controls targeting, delivery, and increasingly creative generation, the advertiser's role narrows to providing a budget and a business objective. Which is exactly what Zuckerberg said the endgame is.
The gap between Meta's vision and marketer comfort is real. 42-44% of consumers say knowing an ad was AI-made doesn't change how they feel about a brand. But that stat cuts both ways — it means more than half of consumers do react differently. And for premium brands where perception matters, that's a material risk.
The strategic question for marketing teams
Meta's automation push creates a fork in the road for every team running paid social.
Path one: full automation. Accept Meta's AI as the primary decision-maker for targeting, creative optimization, and budget allocation. Provide the inputs — creative volume, business objectives, budget — and let the system optimize. This works if your competitive advantage isn't in ad strategy. If you're selling commodity products where the best creative wins on pure performance metrics, letting AI test thousands of combinations faster than any human team can is a legitimate strategy.
Path two: strategic oversight. Use Meta's automation for execution but maintain human control over brand positioning, creative direction, and audience strategy. This requires more effort — you're essentially building a layer of strategic intelligence on top of AI systems — but it preserves the differentiation that comes from understanding your market in ways an algorithm can't.
The teams that will struggle are the ones in between — using automation by default without a deliberate strategy for what human judgment adds to the process.
What this means for marketing investment
Three things change when the platform's AI handles more of the advertising workflow:
Creative strategy becomes the primary lever. When targeting is automated and delivery is AI-optimized, the only input you control is creative. The quality, diversity, and strategic intent behind your creative assets determine campaign performance more than any audience setting.
Brand becomes harder to protect at scale. AI optimization defaults to what performs — which isn't always what's on-brand. Without active oversight, Advantage+ campaigns can drift toward creative variations that generate clicks but dilute brand positioning. The 1,000-asset requirement makes quality control exponentially harder.
Analytics literacy is no longer optional. When the black box is making decisions, the only way to maintain strategic control is through rigorous performance analysis. Understanding what the AI is optimizing for — and whether that aligns with your actual business goals — requires deeper analytical capability than most marketing teams currently have.
The real risk
The risk isn't that Meta's AI performs badly. The data suggests it performs well by the metrics the platform optimizes for. The risk is that marketers confuse platform performance with business performance.
An AI system that improves ad quality by 14% is optimizing for Meta's definition of quality — which is user engagement and relevance within the platform. That may or may not align with your brand strategy, market positioning, or long-term customer value objectives.
The advertisers who treat automation as a strategy will find themselves dependent on a platform's AI with diminishing ability to differentiate. The advertisers who treat automation as a tool — powerful but requiring strategic direction — will use it to execute faster while maintaining the brand and market intelligence that creates lasting competitive advantage.
The bottom line
Meta's Andromeda represents the most significant shift in paid social advertising since the introduction of lookalike audiences. The platform is moving decisively toward full automation, and the technology is delivering real performance improvements.
But technology that works isn't the same as technology that works for you. The marketing teams that thrive in this environment will be the ones that understand what to automate and what to protect — using AI for speed and scale while investing human expertise in strategy, brand, and the kind of market understanding that no algorithm can replicate.
*itscool.ai helps marketing teams build paid social strategies that leverage platform AI without surrendering brand control. If your Advantage+ campaigns are scaling spend but you're losing visibility into what's driving results, the problem isn't the platform — it's the strategy layer on top. Let's talk.*