The problem this solves
Everyone on your team can see competitor ads, yet nobody has time to study them systematically. What actually runs in your category shapes what your buyers expect, and the angles that saturate the feed are exactly the ones your next campaign should route around. But reading five ad libraries, cataloging the claims, and mapping the white space is a full afternoon of tedious work, so it rarely happens, and briefs keep getting written from memory and guesswork instead of evidence.
The cost shows up as creative that blends in. When your ad opens with the same promise as three competitors sitting above it in the feed, you pay for every impression that scrolls past. Concepting against the field takes research plus production, and production is where most teams stall: an angle without a finished image is just a note in a doc. The pack stays theoretical until someone finally books design time weeks out, by which point the field has already rotated.
How the mission runs
- Sweep the competitive field. The mission opens in Ad Libraries, pulling the active Meta and LinkedIn ads for the five competitors you name. The agent catalogs each ad's core claim, offer framing, and visual style, building an evidence-backed picture of what your category's feed actually looks like this month rather than what anyone remembers it looking like.
- Map the white space. With the field cataloged, the agent clusters the claims and isolates the three angles nobody is running: promises competitors leave implicit, objections none of them answer, audiences their framing ignores. Each candidate angle arrives with a one-line rationale tied to the specific ads it is built to counter.
- Write the three concepts. For each angle the agent writes a scroll-stopping hook and supporting body copy in your voice, grounded in your real product pages so every claim stays true. You receive complete, ready-to-run concepts instead of directions that someone on the team still has to sit down and develop.
- Generate six finished images. Image Generation produces a square and a story-format image for each concept, six assets in total. Early passes render at draft quality while composition and subject settle, which keeps iteration inexpensive; once you approve a direction in the thread, the final six render at final quality for delivery.
- Assemble the pack in Notion. The agent builds one Notion page per concept: the competitive rationale, the hook, the body copy, and both image links laid out side by side. The pack reads as a briefing document your team can approve in one sitting, hand to media buying, or feed into a staged campaign.
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 Notion pack containing three complete ad concepts, each with a competitive rationale, a hook, body copy, and two finished images - one square, one story format - six assets in total. Every image is linked at full resolution and rendered at final quality, ready to download and attach to a campaign. The competitor sweep that justified each angle is preserved on the same pages, so the reasoning survives alongside the creative it produced.
Make it yours
- Narrow the sweep to a single dominant competitor and ask for five counter-angles against their current flagship campaign, each concept paired with one finished square image.
- Swap the platforms: read LinkedIn ads only and produce concepts tuned for a B2B feed, with landscape formats replacing the story variants throughout the pack.
- Run it monthly as a standing mission so the pack refreshes as competitors rotate creative, and have the agent flag which of your angles the field has since crowded.
Frequently asked questions
Will the images actually match my brand?
Yes. The agent works from your brand kit - palette, type direction, logo treatment - and applies it across all six images so the pack reads as one campaign. If your kit is thin, the first draft render becomes the reference and you correct direction in the thread before finals are produced.
Can I revise a concept before the finals render?
That is the intended loop. Concepts and draft-quality renders land first; you react in the thread - sharper hook, different subject, warmer palette - and the agent regenerates. Final-quality renders only happen once a direction is settled, which keeps the revision cycle fast and keeps generation spend low.
Does anything publish automatically?
No. The mission ends at an assembled Notion pack with asset links. Nothing touches a live ad account unless you explicitly ask for staging in a follow-up, and even then campaigns are created paused, so a human always presses the final button before any spend begins.