How the ad creative intelligence platform works
Competitor ads are the best free research a media buyer has, and the ad libraries make them almost impossible to use: they vanish when the ad stops, and the links rot. This collects them across every major platform, stores every creative permanently, and layers AI analysis on top so a team can actually work from them.
A look under the hood: why permanence is the hard part, how it understands video and not just images, and how collection, analysis, and briefing live in one owned system.
The short version
It watches competitors' live ads across the major ad platforms, downloads and permanently stores every creative before the platform's own links expire, and applies AI tagging plus deep video analysis on top. The result is a searchable, owned archive a media buyer can actually work from: what's running, what's been running long enough to look like a winner, and what themes and hooks keep showing up. From there it supports the rest of the loop, briefs, swipe-file collections, and a campaign analyzer that reads a performance export and suggests what to kill, keep, or scale.
That's the whole system. The rest of this page is the part that's genuinely hard: keeping the creatives after the ad libraries drop them, and understanding video, not just static images.
It keeps the ads the ad libraries throw away
Public ad libraries are ephemeral. A competitor's ad vanishes from view when they stop running it, and the media links behind it expire on the platform's schedule. This platform downloads every image, video, and thumbnail to durable storage the moment it sees them, deduplicated, so the archive is permanent and owned rather than a set of links that rot.
This is the single most valuable thing it does. Tools that only index or point at ads show you an empty frame a month later. An owned archive lets you study a competitor's whole creative history, including the ads they've since pulled, which is usually the most interesting data.
It understands video, not just images
Every ad gets vision-model tagging (platform, format, themes, likely audience). On demand, a deeper pass handles video properly: it pulls keyframes, transcribes the audio, and analyzes the frames and transcript together, with text extraction across several languages plus translation. Static-image tagging alone misses most of what a video ad is actually doing.
That depth is what turns a pile of downloaded creatives into something a strategist can query: not just "this is a video ad" but the hook, the structure, and the message inside it.
Collection, analysis, and briefing in one owned system
The archive feeds the workflow the agency actually runs: filter and search the creative library, save references into swipe-file collections, and turn those into creative briefs with linked example ads, hooks, and scripts assigned to the creative team. A separate analyzer takes a campaign performance export and returns kill/keep/scale recommendations against real ROAS.
Winner and loser labels are computed from how long an ad has actually been running and whether it's still active, recomputed live, never a stale flag someone forgot to update. It's an honest classification of observed behavior, not a prediction dressed up as one.
Self-hosted and owned, not another per-seat subscription
It consolidates the capabilities of several paid ad-intelligence subscriptions into one self-hosted platform the agency owns outright, on its own infrastructure, at flat cost. The data, the archive, and the analysis all live in-house rather than in a shared community pool or a vendor's account.
For an agency running this across multiple client brands, owning the archive matters as much as the features: the competitive history it builds up is an asset that compounds, and it doesn't disappear when a subscription lapses.
It aggregates public ad-library data into an owned, permanent, AI-analyzed archive. It's built to consolidate what agencies normally rent from several separate tools into one system they control.