6 Research use cases · newest Nov 16, 2026 · all 87 →
AI agent use cases for competitor and customer research.
Real competitor and customer research workflows people are running with AI agents, newest first. Each one links to the original post, lists the tools used, and includes the prompt where it was shared.
Alex P's creative-research workflow: connect the TrendTrack MCP to Claude or Codex and brainstorm problems similar in nature to the one your product solves. Use the MCP to find direct competitors growing fastest or with at least 500 live ads, then run the same search for brands that tackle those similar problems, and pull the top 10 ads for each. Claude analyzes every ad and summarizes the avatars, angles and unique mechanisms, scores each against your product's formulation, and you spin up your own versions where the fit is strongest. Last step: pull up your own proven angle, avatar, mechanism and format combos and see which formats and angles from the search you can apply to them. Alex says he uses TrendTrack himself and the post promotes its Black Friday sale.
Chris Marrano says an agent he built runs every Monday and pulls competitors' live Meta ads. Because the Ad Library hides spend, it sorts ads by how long each has run, on the idea that ads live for weeks are the ones worth studying. It then compares against the previous week: new ads show what competitors are testing, and ads that stopped usually didn't work. A simple, repeatable weekly version of competitor research you can hand to a creative team.
@HenryCrochemore published a Claude skill that takes one domain, pulls the competitor's live Meta account via gethookd, and returns concepts ranked by days running, hooks/angles/offers, the landing page behind each segment, gaps they leave open, and five ads to beat them. His Toniiq example: 279 active ads, 55% dynamic creatives / 36% video, with a "one scoop. all 12 aging pathways" winner live 154 days across 10+ variants. Useful for DTC teams that want a structured competitive teardown instead of scrolling the Ad Library by hand.
Tear down this competitor Meta ads account from the domain I give you. Rank concepts by days running, list hooks, angles, offers and landing pages, then return five ads that attack the gaps.
Have Claude Code build a small app that uses Apify to scrape competitors' Facebook, Google and LinkedIn ads every week and stores the files in S3. Gemini Flash tags what each ad is doing and writes it to a Postgres database. Your team then queries that database from their own Claude Code sessions to find angles for new creative, and a weekly cron loads new competitor ads. The author is building Graphed, which he suggests for hosting it, but the workflow doesn't depend on it.
Jon Loomer has his Muse agent check Meta's Help Center and developer pages for changes every Monday morning. He also used it to build a Value Rules Decision Tool for advertisers.
Meta AI analyzes public content and engagement from similar brands on Facebook and Instagram and tells you where you're ahead or behind and what to change. It checks the brand list with you first.
Benchmark my organic performance against 5 top and comparable brands in my space - am I ahead or behind, what's driving the gap, and what should I change? Check the brand list with me first.