Gumloop vs Airship

Airship has spent this cycle putting AI agents into each authoring surface in turn: a conversational Scene Assistant that builds multi-screen native experiences, then a Campaigns AI Agent that turns a brief into coordinated cross-channel drafts against one shared plan. The newest additions change the job rather than the surface — an accessibility agent that audits a Scene for missing alt text, weak contrast, small type and absent headings and offers fixes in place, and AI Recommendations that reads engagement data and returns prioritised suggestions, shipped alongside the rename of Behavioral Targeting to Audience Pulse. Scenes also gained conditional branching, centralised brand guidelines and saved custom views.

Two things are consolidating. Scenes is becoming the product's centre of gravity: the assistant builds them, branching makes them reactive, brand guidelines constrain them and the accessibility agent checks them. And the agents themselves are widening from generation into review and advice, which is a harder capability to copy than another content generator. The Audience Pulse rename points the analytics side away from individual targeting and toward aggregate engagement, which is the privacy position Airship states directly in the release.

The audit-and-fix pattern is the reusable piece, so the next likely move is applying it beyond accessibility — brand-guideline conformance is the obvious candidate, given the guidelines are now centralised and machine-readable. Timing is unclear from these entries, several of which arrived undated.