The Challenge
A growing herbal brand was posting regularly on Facebook but lacked clarity on what content was actually resonating with their audience.
→ Were they creating content aligned with their Ideal Customer Profile (ICP)?
→ Were they driving engagement or just making noise?
→ What kinds of visuals and messaging were actually working?
The marketing team wanted to be more strategic — but didn’t have time to audit posts manually. They needed something scalable, intelligent, and insight-driven.
The Solution: A Multi-Layered Social Media AI Agent
I built a custom AI automated workflow using layered agents that analyzed both post content and visuals — and provided next-step recommendations based on audience fit and engagement trends.
🔵 Building Strategy Into the System
I started in AI by training ChatGPT to understand my brand voice, business vision, and content goals. I wasn’t just building tools — I was filling in the gaps with intelligence and intuition.
So when this brand came to me, I approached it the same way.
They needed data AND insight — the kind of feedback you’d get from a strategist who understood their ideal customer.
So I designed a custom AI agent workflow that acted like their content strategist: someone who could break down what’s working, what’s not, and how to move forward — without disrupting their current process.
🔵 Scraping With Purpose
First, I built a system that scraped data directly from Facebook using a mix of off-the-shelf and custom Apify actors.
It captured everything: post content, media, screenshots, engagement numbers — no extra manual work required.
🔵Making the Data Think
That raw data flowed into Google Sheets (or Airtable) where it got structured and enriched. I used Google Apps Script and OpenAI to pull meaning from each post — including auto-generated explanations of visuals and video content.
🔵 Layered AI Agents With Human-Like Feedback
Then came the analysis — but not just numbers.
I created two layers of AI:
- The first agent reviewed each post through the lens of the brand’s Ideal Customer Profile (ICP), acting like a mini strategist asking, “Is this who we’re really speaking to?”
- The second agent zoomed out to assess the bigger picture. It reviewed engagement data, the visuals, the ICP notes, and then crafted a human-style summary — with strategic feedback and next steps.
🔵 Insight, Not Overwhelm
What made this system powerful was that it didn’t require new dashboards, extra tools, or added tasks.
Everything ran in the background — fully automated and scheduled to trigger every Monday.
By the time the team started their week, the insights were already waiting for them — analyzed, prioritized, and neatly packaged for decision-making.
They didn’t have to press a button or manage a process. Just open the sheet, see the strategy, and take action.
It was built to think — and to think like someone already on their team.