I built the content operation I would have sold to a client, and pointed it at my own brand: capture, production, voice, memory, scheduling, and measurement, running as one system of 74 skills.
Content Automation
What the system does
- An idea goes from spoken ramble to a filed, structured entry without opening a browser.
- Every draft is scored against a 533-line voice fingerprint and comes back with line-level edits before anything ships.
- Axiom Analytics, a dashboard I built, grades all 327 posts across YouTube, TikTok and Instagram: posts, comments, calendar, ideas and goals in one place.
- Every published piece files into a 179-page vault, so each next piece is written against everything that came before it.
- A health monitor checks every scheduled post twice a day, built after posts died silently on expiring media hosts.
A one-person brand, a full-time build schedule, and no team behind it.
Most of my time goes to building, and I didn’t want content creation eating into that. So I started building out skills one by one to make the process more efficient: thumbnail generation, scripting, recording, motion graphics, video editing, cutting long-forms down into shorts. It all slowly stacked up. Now 74 skills make the whole process way quicker.
Four systems, one compounding engine.
Video production
Motion graphics render onto my talking-head footage, word-synced straight from the transcript. Agents scan each long-form for short-worthy moments, fan out one per short with the animations added, and wait for my review. Thumbnails and carousel art get generated through image-model MCP servers (Higgsfield, Nano Banana, GPT Image) with my own best performers as the style references.
Memory and voice
Every script and video I’ve ever made is documented and studied: my structure, my tone, the way I actually talk. So when I ramble out a raw idea, it comes back as a script that sounds like me, and the same brain writes my captions, descriptions, carousels and comment replies. It’s a second brain of me, purely for content.
Ship and monitor
Posts schedule out to three channels through Buffer’s API. A monitor checks every scheduled post twice a day and pings me on Telegram if anything dies. It’s also wired into Meta through an MCP server, so comments can trigger automations: someone comments “sheet” and the system DMs it to them automatically.
Axiom Analytics, the measurement layer
A reporting app I built to grade my own work: posts, comments, calendar, ideas and goals in one place, synced from three platform APIs plus Buffer. Every post gets scored against the trailing median of its own platform and format, so a win means beating my own baseline, not an invented one. The insight cards tell me what to double down on and what to drop, computed from my own history.
Claude Code
FFmpeg
Playwright
Next.js
Supabase
Netlify
Buffer
Notion
Obsidian
Telegram
Curious what a system like this would look like pointed at what you’re building?
Reach out.
Work With Me