How I Built an AI Competitive Ads Agent That Scans Meta, LinkedIn, and Google in Minutes
Every marketing leader I know does some version of competitive research. They check what competitors are running on LinkedIn. They scroll through the Meta Ad Library. Maybe they pull up Google Ads Transparency Center once a quarter. It takes hours. It is usually incomplete. And by the time the findings make it into a deck, the data is already stale.
I got tired of that cycle. So I built an AI agent that does the entire scan in minutes.
What the competitive ads agent actually does
This is not a prompt that summarizes a competitor's website. This is a full competitive intelligence system built with Claude Code that automates what a junior analyst would spend a week doing.
Here is the workflow:
Step 1: You give it a list of competitors. That is it. Just the names.
Step 2: The agent opens Meta Ad Library, LinkedIn Ad Library, and Google Ads Transparency Center. It navigates each platform using browser automation, pulls live ad creative, and takes screenshots of everything it finds.
Step 3: It extracts the data that matters. Not just "here are their ads." It pulls headline patterns, pain points being targeted, value propositions, CTAs, creative formats, and estimated spend levels. It organizes all of this into structured tables.
Step 4: It finds the patterns. This is the part that used to require a senior strategist. The agent identifies three messaging patterns that are working across your competitive set, three gaps that nobody is filling, and the top five ads worth studying with notes on how to adapt them for your brand.
The entire scan runs in minutes. Not days. Not weeks. Minutes.
Five ways to use it
The agent supports five different workflows depending on what you need:
Full scan. The default. Complete competitive ad audit across all three platforms with the full analysis. This is what you run quarterly or when entering a new market.
Campaign planning. You are about to launch a campaign and want to see what messaging is already in market. The agent extracts patterns and delivers a brief with three ad variations you can build from.
Positioning research. You want to understand where competitors are clustering their messaging and where the white space is. The agent maps the competitive positioning landscape and shows you where to plant your flag.
Creative inspiration. You need fresh ad formats, hooks, and CTAs. The agent pulls the most creative examples from your competitive set and adapts them to your brand voice.
Trend tracking. Run it monthly to catch shifts in competitor messaging before they become obvious. This is how you stay ahead instead of reacting.
What makes this different from just using ChatGPT
Three things.
It actually visits the ad platforms. It does not summarize what it thinks competitors might be running based on training data. It opens a browser, navigates to Meta Ad Library, LinkedIn Ad Library, and Google Ads Transparency Center, and pulls what is live right now. Real data, not guesses.
It knows your brand. The agent reads your brand voice rules, your personas, and your positioning before it runs. So every insight is framed in terms of what matters to your buyers, not generic competitive analysis.
It saves everything. Screenshots, ad copy, analysis tables. All organized in an output folder with the date and competitor names. You have a competitive intelligence archive that builds over time.
The results
I have run this agent for my own business and for clients. Here is what it replaces:
- A junior analyst spending 8 to 10 hours compiling a competitive ad audit. The agent does it in under 15 minutes.
- An agency charging $3,000 to $5,000 for a quarterly competitive review. The agent runs on demand for the cost of the API call.
- The "I know what competitors are doing because I saw an ad last week" approach that most marketing teams actually rely on. The agent gives you systematic coverage, not anecdotal awareness.
Try it yourself
I open-sourced this agent. You can install it as a Claude Code skill and run it against your own competitors today. No custom setup required beyond having Claude Code and the Playwright browser tool.
The public repo is at github.com/Laura-Beaulieu/growthloops-claude-skills. The competitive ads agent is one of three skills available, along with a content repurposing engine and a full marketing strategy audit.
Clone the repo, add your brand details to the CLAUDE.md file, and run /competitive-ads with your competitors' names. That is it.
What this looks like at scale
This agent is one of 20 that I have built as part of the GrowthLoops marketing operating system. Each one handles a specific function that used to require a dedicated person or an expensive agency.
The competitive ads agent is a good place to start because the value is immediate and obvious. You run it once, you see what your competitors are doing, and you have actionable intelligence in minutes.
But the real power comes when you start chaining agents together. Run the competitive scan, feed the gaps into the content repurposing engine, and you have a full content response to competitor positioning in an afternoon. That is what a go-to-market engineering workflow looks like in practice.
If you are a B2B SaaS marketing team that wants to see what building your own AI agents looks like, start with this one. It is free, it is open source, and it works.