Runway Ads closes the loop between creative and performance

Runway has launched Runway Ads, an agent that generates paid ad creative, sends approved variants to Meta, Google, and TikTok, and uses campaign results to produce the next batch. The system extends Runway beyond video and image generation into campaign execution and measurement.

Runway has used the product for its own paid marketing since July and is opening early access to enterprise partners. A broader rollout is planned in the coming weeks, although the company has yet to disclose pricing, firm availability dates, or API access.

From brand kit to live campaign

Runway describes the product as an autonomous advertising system, although human approval is enabled by default. In this context, an agent is software that chains together creative production, publishing, measurement, and revision without requiring a person to move assets and data between tools.

1. Ingest: The agent reads brand guidelines, product images, previous ads, and campaign instructions.
2. Generate: It creates video and image variants, including different hooks, layouts, copy, and formats.
3. Review: Automated brand checks run before the assets enter an approval queue.
4. Publish: Approved ads are sent to connected Meta, Google, and TikTok accounts.
5. Measure: The system retrieves delivery and performance data from each platform.
6. Regenerate: Results from the campaign guide the next set of variants.

The feedback loop is the product’s central feature: performance data becomes an input for subsequent creative. Teams that currently export reports, brief designers, resize assets, and upload new ads could consolidate those steps into one workflow.

Controls for brands and budgets

Runway Ads includes localization and campaign controls intended to make large-scale generation manageable. The system can update on-screen text, product screenshots, and voiceovers for different markets, then resize each asset for the required channel and aspect ratio.

These controls matter because high creative volume can overwhelm manual review. A team producing hundreds of variants each week must decide which changes require inspection, which can rely on automated checks, and which campaigns may publish without individual approval.

Runway’s own results

Runway built the system to address constraints in its growth marketing workflow. The company reports the following results from using Runway Ads on its own campaigns:

Those figures come from Runway’s own funnel and have not been independently verified. The company has also not published its attribution method, test design, campaign mix, or the period used to calculate the claimed $100 million in annual recurring revenue. The results therefore show how the system performed for its maker, rather than a benchmark other advertisers can assume.

Who Runway is targeting

Runway Ads builds on Runway Agent, which already handled briefs, asset generation, and campaign analysis through a conversational interface. The new product adds platform publishing, performance ingestion, and budget guardrails for several customer groups:

• Global enterprises that need consistent localization and testing across many markets.
• Ecommerce companies that can generate campaigns from large product catalogs.
• Small marketing teams that need more creative variants without expanding production staff at the same rate.
• Agencies that want a repeatable generation and testing workflow across client accounts.

AI development platform Lovable is the launch partner and is using the product to expand its own advertising output.

The contest for campaign data

Meta and other advertising platforms already offer generative creative tools within their campaign managers, while ad networks such as Taboola have added automated creative and campaign features. Runway’s approach spans Meta, Google, and TikTok while using its own image and video models.

Cross-platform access could give Runway a broader view of creative performance, provided its integrations expose consistent and timely data. Each platform uses different campaign structures, attribution systems, optimization events, and reporting delays, so a metric labeled “conversion” may not represent the same event everywhere.

The feedback data also has strategic value. As generation costs fall and advertisers produce more variants, campaign results can reveal which messages, formats, and visual treatments work for particular audiences. Platforms benefit from keeping those signals inside their own systems; Runway is attempting to coordinate them across competing channels.

Questions the rollout must answer

• Integration depth: Runway has not detailed which campaign types, bidding modes, conversion events, or reporting fields each platform supports.
• Attribution: The company has not explained how it reconciles conflicting conversion claims or attribution windows across platforms.
• Developer access: Public documentation does not yet specify APIs, webhooks, export formats, rate limits, or integration options beyond the product interface.
• Data handling: Advertisers will need details about retention, model training, account permissions, regional storage, and access controls.
• Creative evaluation: Higher output does not establish whether the system can isolate why an ad performed well or avoid overfitting to short-term results.
• Operational review: Teams producing hundreds of variants will need sampling, audit logs, version history, and rollback controls to supervise automated publishing.

Runway Ads packages generation, deployment, and measurement into a single system, reducing the manual handoffs that slow performance marketing. Its broader value will depend on the quality of its platform integrations, the reliability of its attribution, and whether advertisers can supervise high-volume automation without creating a new review bottleneck.