This is one of the most recurring questions we're getting right now. Here's what you need to know.

The big picture

Shifts defining paid media in 2026:

1. Creative is the new targeting. Meta’s Andromeda algorithm controls targeting now. Creative diversity and volume are the only real levers. Paid social is becoming 80% creative ops, 20% media buying.
2. Competitive moat = proprietary data + feedback loops. Sure, creative is the new targeting. That's been known for a while. The real advantage, though, isn't making more creatives. The winners will be those who have (or find a way to generate) proprietary data and the tightest learning loops.
3. The media buyer is becoming an AI strategist. The job is shifting from manual optimization to AI systems architecture. IAB Lab released an agentic AI roadmap. Platforms are building autonomous buying infrastructure. The skill isn't disappearing, but it is shifting from "manage campaigns" to "architect the system."
4. Campaign architecture is collapsing. Advertisers who consolidated to 2-campaign structures saw 32% CPA drops.
5. Organic content now feeds the paid algorithm. Organic posts, Reels, and videos now directly influence ad ranking. The boundary between paid and organic has dissolved.
6. Ads are coming to ChatGPT. OpenAI is testing ads inside ChatGPT as of this month. An entirely new channel is forming.
7. The entire ad engine was rebuilt. Andromeda + GEM didn’t update the old system; they replaced it.

Inside look

• Taylor Holiday built Compass — a system that calculates monthly ad volume from financial models, then uses AI to generate creative angles from 30 psychological purchase drivers. The most sophisticated operator-built AI infrastructure we’ve seen described publicly.
• Dara Denney encoded her creative analysis framework into an AI agent. Pick any ad in your account and it diagnoses performance against her proven framework. Systems that scale your judgment, not replace it.
• FULLBEAUTY Brands used Meta’s native Advantage+ Creative AI to replace plain white catalog backgrounds — 45% higher ROAS, 22% higher conversion rate, 36% higher CTR. One of the clearest wins for Meta’s built-in AI tools.
• Jon Loomer built JonBot — an AI trained on 700+ pages of his content. He shares where it speeds him up, where it fails, and the guardrails he’s built. Honest and practical.
• ​Motion surveyed 500+ advertisers: 86% plan to increase AI use for creative ideation, but 55% aren’t convinced by output quality. Nielsen found consumers find AI-generated ads “annoying, boring, and confusing.”

Frontier players

Operators doing interesting work at the intersection of AI + Meta Ads.

• ​Cody Plofker — CEO of Jones Road Beauty. Writes weekly with real numbers from a brand he actually runs.
• ​Dara Denney — $100M+ in Meta spend. Building AI agents for creative analysis.
• ​Taylor Holiday — CEO of CTC. Analytically rigorous. Will challenge your ROAS assumptions.
• ​Andrew Foxwell — Runs Foxwell Founders (500+ advertisers, $300M+/mo in Meta spend).
• ​Jon Loomer — Old school, yet still cutting-edge.
• ​Shanik Patel — Ex-Grammarly. Fascinating work with AI + video ads.
• ​Barry Hott — $600M+ managed spend. His “ugly ads” thesis is what Andromeda actually rewards: authenticity > polish.
• ​Sarah Levinger — Living at the psychology × creative × AI intersection. Works with Hexclad, True Classic, Fabletics.
• ​Rob Leathern — Former VP Product at Facebook. He literally built the systems that govern how Meta ads work.

The stack

AI tools for Meta ads.

Creative Generation

• ​Arcads — AI UGC video ads. $16M seed (Sequoia Scout). 100K videos/month. Hit profitability with 7 employees.
• ​Creatify — End-to-end AI video ads from a product URL. New “AdMax” combines creation + testing + analytics. $15.5M Series A, $9M ARR in 18 months.
• ​MakeUGC — Scripts to AI UGC with 300+ avatars and product-in-hand demos. Most affordable entry ($49/mo).
• ​Mintly — AI ad creative studio. Templates reverse-engineered from winning ads. Upload product photo → remade in top-performing formats.
• ​Pencil — AI ad creation with predicted performance scores trained on $1B+ in spend.
• ​Runway — AI video generation at cinematic quality for brand-level creative.

Campaign Optimization

• ​AdAmigo — Autonomous AI media buyer for Meta. Monitors 24/7, catches anomalies, executes via text/voice. $99/mo. Closest thing to replacing a junior media buyer.
• ​Koast — Bulk Meta ads launching (30 at once), automated stop-losses, multi-account sync. Built for agencies, not marketers.
• ​AdStellar — 7 specialized AI agents build complete Meta campaigns in 60 seconds. Multi-agent architecture.
• ​Madgicx — AI Meta ads management. Auto budget reallocation + creative fatigue alerts.

Creative Intelligence

• ​Segwise — Multimodal AI auto-tags every creative element and maps to performance. Only platform doing element-level analysis.
• ​Superads — Chat with your ad data. Auto-categorizes winning hooks, triggers, copy patterns. Free tier. By Superside.
• ​AdSkate — Tests ads BEFORE launch via synthetic audience models. Lidl: 24% CTR boost.
• ​Replai — Computer-vision AI auto-tags video ads at frame level. Sequoia-backed. Processes $5B+ in spend.
• ​Motion — Creative analytics. Which ads are winning and why.
• ​Foreplay — Competitor ad tracking + AI creative briefs.

Competitive Intelligence

• ​Panoramata — Tracks competitors’ full stack: Meta ads, emails, landing pages, SMS, website changes. 4M+ ads searchable. See the full funnel.

Speaking of staying on the frontier... a quick message from our sponsor, Gather.

What AI tactics are actually useful in your marketing and growth efforts?
​
Join your peers in San Francisco for a candid conversation about what's working, what's not working, and what the future of AI in marketing/growth looks like in 2026.
​
Food and drinks will be provided. Limited availability so reserve your spot today!

Signal vs Noise

💨 “We can replace our media team with AI.” AI handles execution (bids, placements, audiences). Someone still needs to architect the strategy, direct the creative, and interpret what’s working. The teams massively cutting headcount are about to learn this the hard way.

🎯 The media buyer role is evolving into an AI strategy architect.

💨 Using AI to generate ad copy and calling it a strategy.

🎯 Using AI upstream — customer research, persona development, creative briefing.

💨 More creative volume = better results. Cody Plofker disagrees: Meta classifies visually similar ads as the “same ad” even with different hooks. His team cut from 4-5 variations per ad to 2-3. Variation testing is “a big cost center and largely wasteful.”

🎯 Creative diversity across angles and formats. Not iterations of the same concept.

💨 Platform-reported ROAS. Every platform’s AI is optimized to make itself look good.

🎯 Incremental Attribution and holdout tests. Your number gets smaller but (a little more) honest.

Rabbit hole

Want to go deeper? Start here:

• Alvin Ding on how Meta’s AI might be damaging your brand — AI slapping music on videos, cropping assets, misrepresenting brand voice. The dark side of Advantage+ without oversight. 📕
• Monica Shukla in AdExchanger on what Andromeda actually changes and what it doesn’t — the most grounded strategic take: brands seeing lift “adjusted not by outsourcing judgment to the algorithm but by refining their creative strategy.” 📕
• Andrew Foxwell on Perpetual Traffic: The Secret to Meta’s Andromeda Revealed (Part 1) and Part 2. From someone running a community spending $500M/mo on Meta. 🎧
• Why you can’t rely purely on direct conversions when measuring Meta Ads success 🎧

Whatcha think?

This is the first edition of The Frontier — a new format we’re experimenting with. We’ll be running this for the next few weeks as we dial it in.

Love it? Hate it? Want different topics? Hit reply. We read 'em all.