Runway's new AI engine makes, tests, and buys ads autonomously

Performance marketing has never really been about finding one perfect ad. It is an endurance sport of feeding the algorithm enough creative variants until something clicks.

Runway just launched Runway Ads, an autonomous AI engine designed to run paid ad creative from prompt to publish.

Closing the loop

Instead of treating AI video generation as a standalone design step, Runway Ads wraps the entire creative workflow into a self-optimizing loop.

Marketers connect their ad accounts and brand kits. The platform then generates image and video creative, pushes approved assets directly to Meta, Google, and TikTok, reads back performance data, and generates the next wave based on what actually spent money.

It also handles multi-language localization on its own. Rather than just swapping out a voiceover track, the engine translates on-screen text and updates in-app UI screenshots while respecting brand voice rules.

Human approval is kept on by default, and every variant goes through an automated brand check before going live, though teams can toggle to full auto once they trust the outputs.

Internal testing wins

Runway built this tool primarily to solve its own marketing bottlenecks, according to co-CEO Anastasis Germanidis.

Built on top of Runway Agent, the company has been using the setup internally since July to scale its own growth campaigns on Meta and TikTok.

The reported metrics are striking. Runway's weekly ad creative volume jumped from 77 to roughly 900 ads per week, while return on ad spend doubled.

Conversion rates rose by roughly 34% with click-through rates remaining steady, driving cost per subscriber down 41% even as total spending grew. Early access is now open on Runway's website.

Why it matters

For growth teams and lean startups, this fundamentally changes the math on creative production. The bottleneck in paid social has rarely been analytics, it has been the human bandwidth required to cut, resize, and localize fifty variations of a single concept. Automating that grind lets small teams test like enterprise giants.

Strategically, Runway is making a brilliant pivot from generic creative software to performance ad spend, which is where real enterprise budgets live. By owning the feedback loop between generation and distribution, Runway avoids being treated as a commoditized video model vendor.

Still, we think this opens a real risk of creative fatigue. When every brand can flood feeds with hundreds of machine-optimized variations, audience blindness will accelerate. Optimizing purely for whatever grabs spend could easily tilt ad creative toward repetitive clickbait unless teams keep a tight leash on quality.