The problem this solves
"Make it feel like this" is how creative direction actually happens. Someone finds an ad that works - the pacing is right, the hook lands, the text placement feels premium - and forwards it with that sentence. What follows is the lossy part: the recipient watches it, forms their own impression of why it works, and produces something that resembles the reference the way a retelling resembles a story. The specific decisions that made the original work were never written down, so they never survived the handoff.
Decoding a reference properly is real analytical work. What happens in the first second versus the third? When does text first appear and what register is it in? How many cuts, how long is each beat, where does the product enter, what does the end card say? These are answerable questions with exact answers, and almost nobody answers them before briefing, because extracting them means scrubbing through the video frame by frame with a notebook.
This mission does the decode and the translation. It analyzes the reference - hook, storyboard, pacing, every piece of on-screen text quoted - then writes the brief that transposes those mechanics onto your brand and generates concept frames so the direction is visible rather than described.
How the mission runs
- Decode the reference. The video is analyzed into its mechanics: a burst across the opening seconds, key frames across every beat, the transcript, and all on-screen text exactly as rendered. The output is the reference as a document instead of an impression.
- Name why it works. From the storyboard the mission extracts the transferable decisions: the hook formula, the beat structure and cut rhythm, when and how text appears, where the product enters, how the CTA is staged. Style is separated from substance, so what transfers is the craft rather than the competitor's branding.
- Write the brief for your brand. The mechanics become a creative brief in your workspace's own context: your product in the product slot, your proof in the proof slot, your voice in the copy lines, with the reference's timing and structure preserved and cited beat by beat.
- Generate the concept frames. Two concept frames are generated to set the visual direction - the hook moment and the product moment as your brand would shoot them. Frames anchor the conversation; nobody argues about adjectives when there is an image.
- Deliver the package. The reference decode, the brief, and the concept frames land together, ready for your creative team or for the generation playbooks to take further. The reference's key frames ride along, so the comparison is always one glance away.
The prompt
This is the exact objective the agent receives. Swap the obvious placeholders for your own domain, segment or channel and run it as-is from the console, Slack, or the API.
What comes back
A creative package built from the reference: the decoded storyboard with quoted on-screen text and timing, a brief that transposes the working mechanics onto your brand beat by beat, and two generated concept frames setting the visual direction. "Make it feel like this" becomes a document a team can actually execute.
Make it yours
- Feed it your own best-performing past ad as the reference, so the brief extracts and reuses what already worked for you.
- Run three references through it and ask for the common mechanics, turning taste into a named house style.
- Send the brief straight into the video-generation playbooks, so the concept frames become a full generated draft cut.
Frequently asked questions
What can be used as a reference?
Any direct video URL or Loom link, a competitor creative pulled from the ad libraries, or a file attached in chat. If it plays, it can be decoded.
Is building from a competitor's ad acceptable?
Mechanics are not property: pacing, structure, and hook formats are the shared craft of advertising, and every strong creative team studies them. The brief transfers the decisions while replacing every branded element - your product, your claims, your voice - which is exactly how references have always been used, now with the decode written down.
Why concept frames instead of a full generated video?
Because direction should be agreed before production is spent. Two frames settle the visual argument cheaply; the full cut is one playbook away once the direction holds. Teams that skip to production iterate on the expensive step instead of the cheap one.