Lessonpills is an educational platform providing high-quality, instant English lessons (video, audio, or article-based) structured around renowned publications. This project proves I can design, build, and maintain an agentic content engine that generates, publishes, and promotes specialized content at scale.
1. Agentic lesson plan generation (from prompt to publication)
The core challenge of Lessonpills was maintaining high instructional quality while achieving complete automation. I solved this by building an advanced, structured workflow where an AI agent (Google Gemini/Claude) follows meticulous, structured prompts referencing specific, high-authority textbooks (e.g., Oxford and Cambridge).
| Workflow Step | Make.com Modules Used | Key Function & Output |
| Input & Database | Notion (Watch Database Items) | The process starts when an idea or topic is tagged as "Ready" in the Notion database. |
| AI Generation | Google Gemini/OpenAI (Claude) | The AI agent receives a detailed prompt (including tone, structure, and academic references) and generates the full lesson text, quiz questions, and metadata in a structured JSON format. |
| Content Formatting | Data Stores, JSON, HTTP | The raw JSON output is parsed, validated, and converted into Ghost CMS-compatible Markdown (HTML/MD). |
| Image Creation | Renderform (or similar graphics module) | I automated the creation of custom, branded lesson thumbnails and social media graphics by dynamically injecting data (lesson title, level) into a template. |
| Publication | Ghost CMS (Create a Story), Cloudinary (Upload) | The final, formatted content and graphics are automatically uploaded, scheduled, and published on www.lessonpills.com. |
This scenario is exemplified in the logic of the Lessonpills Notion database > Ghost blueprint, which handles content extraction, transformation, and final CMS delivery.
2. Automated social media distribution
After the lesson is published, a separate automation scenario takes over to guarantee maximum visibility across all primary social channels. This eliminates the manual, repetitive task of repurposing and posting.
The Lessonpills social Ghost scenario watches the Ghost CMS for new publications and executes the distribution logic:
| Social Media Module | Channel & Purpose | Output Format |
| Ghost CMS (Watch Posts) | Source Trigger | Initiates the workflow upon new post. |
| Renderform/Cloudinary | Asset Generation | Pulls the lesson details to create unique images/videos for each channel's dimensions. |
| OpenAI/Claude | Copywriting Agent | Generates tailored, platform-specific copy (e.g., short, engaging Instagram copy vs. a professional LinkedIn insight post). |
| Bluesky/Facebook/LinkedIn/Instagram | Distribution | Posts the unique copy and graphics, tagging the lesson URL. |
This automation ensures the message is customized for the audience of LinkedIn, Facebook, Instagram, and Blueskyâall automatically, based on a single source of truth in the Notion database.
3. Video content generation (Quiz Shorts)
To drive traffic from platforms like YouTube and Instagram Reels, I built dedicated, high-speed video creation automations that do not rely on a full lesson being published.
These scenarios generate engaging, rapid-fire grammar quizzes and instantly push them to YouTube Shorts and Instagram Reels.
A. Grammar Test Shorts (Manual/API Trigger)
This blueprint demonstrates rapid, on-demand content generation for quick-hit videos.
| Make.com Modules Used | Key Function & Output |
| Webhooks/Custom Input | Trigger |
| OpenAI (GPT-4 Mini) | Video Scripting Agent |
| Renderform/Descript | Video Creation |
| YouTube, Instagram, Facebook | Distribution |
B. Grammar Test Long Videos (Notion Triggered)
This scenario is designed for more structured, evergreen video content integrated directly with the Notion database.
| Make.com Modules Used | Key Function & Output |
| Notion (Watch Database Items) | Source Trigger |
| OpenAI/Claude | Script & Narration |
| Renderform/Descript | Production |
| YouTube | Distribution |
This architecture showcases my expertise in using AI agents for structured data output (JSON), managing complex flow logic, and integrating with advanced creative tools like Renderform and Descript to fully automate high-volume content production.
Content collaboration isn't on your agenda at the moment? Let's stay connected on LinkedIn for your future projects. đ¤
