Skip to content
AI Primer
release

Runway opens Ads engine to select enterprise partners

Runway says select enterprise partners can now access its Ads engine. In its own use, the company says the engine increased ad volume 1,000% and doubled return on ad spend; wider rollout is planned in the coming weeks.

4 min read
Runway opens Ads engine to select enterprise partners
Runway opens Ads engine to select enterprise partners

TL;DR

  • Runway Ads connects an ad account and brand kit to a loop that generates creative, publishes approved variants, and learns from spend on Meta, Google, and TikTok, according to Runway Ads launch post.
  • Runway says its own weekly ad volume rose from 77 to roughly 900, while ROAS doubled, conversion rose 34%, and cost per subscriber fell 41%, with CTR steady, according to Runway's results post.
  • Human approval is on by default, with automated brand checks, localization, resizing, and budget limits described in Runway Ads launch post.
  • Select enterprise partners, including Lovable, are piloting the product now, with a wider rollout planned in the coming weeks, Runway Ads launch post says.

Runway's launch post includes a full handoff from generated asset to live campaign, with performance data feeding the next creative batch. The product page exposes the loop as Generate, Approve, Publish, Learn, while an independent summary identifies Lovable as the launch partner.

Generate, approve, publish, learn

Runway frames Ads as the first step toward automating commercial workflows while scaling creative production, [iamneubert's announcement] iamneubert's announcement says. At Runway's summit, Runway's stage demo showed CPO Anthony Maggio introducing the product interface.

The product page lays out four stages:

  1. Generate: Runway's product page says Ads uses brand guidelines, product imagery, and past winning ads to create video and image variants, tagged by hook, audience, format, and call to action.
  2. Approve: An automated brand check runs before the approval queue, with human approval enabled by default, according to the launch post.
  3. Publish: Approved variants go directly to Meta, Google, and TikTok, the product page says.
  4. Learn: The launch post says performance data returns from those platforms and guides the next round toward the combinations that earn spend.

Controls and localization

Runway Ads localizes on-screen text and product screenshots alongside voiceover, following rules for brand voice and terms that should not be translated, according to the product page. It also reformats each variant for different aspect ratios and channel specifications.

Budget movement is configurable. Runway says teams can set a maximum daily change, a minimum share of net-new audience, and retargeting caps, with the system staying inside those limits.

The product's boundary is explicit in iamneubert's clarification: customers bring their campaigns and manage performance. Runway is automating the creative production, publishing, measurement, and iteration loop around those campaigns.

Runway's own program

Runway built Ads on its own performance marketing infrastructure, particularly for Meta and TikTok. The company presents these as internal results from its program in the launch post:

  • Ad volume: 77 to roughly 900 ads per week.
  • Return on ad spend: approximately doubled.
  • Conversion: up roughly 34%, while click-through rate held steady.
  • Cost per subscriber: down 41%.
  • Annual recurring revenue: Runway's social post attributes $100 million in ARR to the performance marketing effort, Runway's results post says.

Runway co-CEO Anastasis Germanidis said consolidating ad generation, publishing, and measurement gave the company visibility into paid advertising efficacy and helped it move faster, according to the launch post.

Asset-level workflows

The evidence around the launch also shows how individual creators are assembling ad assets before a campaign system takes over. A two-hour example used Opus 5.5 and Magnific MCP, as techhalla's workflow describes.

That workflow included prompts for multiple models and a separate editing pass. For expensive 30-second Seedance 2.5 generations, techhalla's draft-mode note recommends testing in draft mode before moving a selected result to HD.

Runway Ads operates one layer above those asset recipes. Its inputs are a brand kit, product imagery, past ads, and connected platform data, which it uses to create and test variants rather than only produce a finished spot, according to the product page.

Human review at scale

The automation leaves a monitoring problem in view. tomaldertweets' warning argues that 400 creatives can burn budget for ten days before anyone notices, because human attention does not scale with output volume.

Runway's default approval queue and configurable spend limits are the product's stated controls for that operating environment, as described in the launch post.

Access and availability

Runway says it is piloting Ads with select enterprise partners, including Lovable, before a wider rollout in the coming weeks. The product page directs interested companies to an early-access form, after which Runway's Enterprise Sales team will make contact.

Further reading

Discussion across the web

Where this story is being discussed, in original context.

On X· 4 threads
Generate, approve, publish, learn2 posts
Controls and localization1 post
Asset-level workflows2 posts
Human review at scale1 post
Share on X