A performance marketer’s entire content operation — blog, video scripts, competitive intelligence, and self-auditing infrastructure — run by autonomous AI agents. Every agent, every API, every dollar.
This is the live view of my autonomous content pipeline — 6 AI agents built on Claude Code that handle everything from SEO blog writing to Instagram script generation to competitive intelligence. The entire system runs on performance marketing automation principles: trigger-based execution, centralized reporting through Telegram, and a self-auditing Meta advertising workflow that catches drift before it becomes a problem. Total cost: $80/month.
What Fires, When, and What It Costs
The SEO Validator
Last line of defense for SEO. Weekly deterministic check of every published post and the homepage — 12-point HTML checklist + 6 DNS health checks + cross-site dedup checks. Auto-fixes OG tags, SVG aria-labels, and stale sitemaps. Verifies rendered HTML, not just API/DB values.
The Sitemap Freshness Monitor
Twice-daily early warning for sitemap staleness. Compares published post count to sitemap entries and auto-fixes mismatches via 3-layer cache flush + regeneration. Prevents new posts being invisible to Google for hours.
The Content Analyst
Instagram Reels analytics agent and platform. Scrapes @neimansview via Apify, computes hook/length/trend patterns, generates action items, runs Gemini 2.5 Pro video analysis. Dashboard at localhost:8500 with pre-publish video scoring. Potential standalone product for creators.
The Auditor
Reviews all employees weekly for drift. Runs voice learning loop (compares Original Draft vs Final Version), checks research brief coverage, proposes voice-calibration.md updates.
Git Sync
Auto-commits and pushes ~/.claude and agent-tracker repos to GitHub on weekdays.
Dream (Memory)
Validates, prunes, and reorganizes all memory files weekdays. Keeps the system’s knowledge clean and current.
Where the Money Goes
Per-post cost if Blog Writer runs M/W/F = $0
How the Agents Talk to Each Other
What the Agents Are Saying Right Now
Why I Built This
I got tired of the bottleneck being me. I manage eight-figure ad budgets for DTC brands, and the content side of the business — blog posts, video scripts, competitive intel — was always the thing that slipped. Not because it wasn't important, but because there are only so many hours in the day.
So I built an autonomous content pipeline using Claude Code as the runtime for every agent. The whole system talks through Telegram as a command bus, stores state in Notion, and publishes directly to WordPress. There's no dashboard, no SaaS platform in the middle. Just config files, scheduled jobs, and AI agents that audit each other so I don't have to.
Here's the thing: this isn't a demo. These agents run my actual performance marketing content operation right now. The Blog Writer publishes three posts a week. The SEO Monitor and Optimizer keep rankings climbing. The Dream agent consolidates memory every weekday. Total cost: less than what most people spend on a single freelance blog post.
I'm building in public because I think more marketers should see what's possible when you stop treating AI like a chatbot and start treating it like infrastructure.