Social Rocket
Runway now clearly positions Agent 2.0 as a marketing tool: the AI agent is meant to analyse campaigns, generate creatives and adapt existing content for further formats and markets.
The foundation for this isn't new: Runway introduced Agent 2.0 on 25 June 2026. What stands out, however, is how clearly the company is now tailoring the agent to real marketing work. In its official product description, Runway explicitly names campaign data from Meta, YouTube, TikTok and Google as possible inputs for analysis.
From video generation to marketing workflow
The original Runway Agent was primarily a dialogue-based creative partner for video. Agent 2.0 extends this approach with marketing context.
According to Runway, teams can provide existing creatives and performance data. The agent is meant to recognise which patterns stand out in a campaign and then generate new content, variants and formats from them. For social content, Runway mentions automatic crops for 9:16, 16:9 and 1:1.
Generative video AI is thus shifting from single production toward an ongoing workflow of analysis, production and scaling.
Creative intelligence becomes its own layer
This development fits a broader trend in the advertising market. Kroger recently showed what AI-based creative scoring with real conversion data can look like. At the same time, Meta is expanding its AI features for creatives and attribution.
Automation in the advertising business is also increasing. Our overview of AI advertising and automated ad spend shows that models increasingly do more than generate content: they also support decisions in the campaign process.
Runway addresses a different point than the media platforms themselves: the agent sits within the creative workflow and connects briefing, asset production and the evaluation of existing campaign signals.
More variants, higher demands on measurement
For marketing teams, the most relevant point is the higher production speed. A starting creative can be turned into multiple formats and markets without rebuilding each step separately.
At the same time, this raises the demands on clean measurement. When many variants are created automatically, teams need to be able to clearly separate which changes actually correlate with better results and which merely add production volume.
Runway itself frames the ambition broadly: Agent 2.0 is meant not only to deliver individual assets, but to support marketing work across several steps.
Not yet a fully autonomous campaign manager
The limits of the current product are important. Runway writes itself that the agent should, in the future, connect more directly with platforms and learn from performance data. This vision therefore partly describes further product development.
Today, Agent 2.0 is primarily a shared interface for analysis, creative production, formatting and localization. The actual media steering must be considered separately.
For social media and performance teams this is still relevant: the creative process is becoming increasingly agentic, while measurement and approval must remain structured.
Sources
Runway: Agent – Create, analyze and scale your marketing
Adweek: Runway’s AI Video Agent Creates and Measures Ads Across Platforms