42 Articles in 14 Days: Building a Content Factory That Never Sleeps

The Spark

I noticed I was writing the same three types of articles every day - Time Travel guides, Cold Cases, and Hollywood vs History breakdowns. Each one took 30-45 minutes of focused work. Multiply that by three, and half my day vanished before I touched anything else.

The question wasn't whether to automate. It was how to do it without the output feeling like slop.

What shipped

The main arc: Content pipeline at scale

• 42 articles published across historiqly-blog in 14 days (Time Travel, Cold Cases, vs Hollywood)
• Tweet syncs now auto-commit daily to keep social and blog aligned
• Three parallel digest streams running on cyril.page - Science, AI Video, and Daily

Branding overhaul

• Switched Historiqly's primary color from blue to warm brown-gold
• Redesigned landing page: rotating taglines, flanking portraits, new hero section
• Generated fresh favicon set (16x16 through 1024x1024)
• Auth forms now match the warm dark theme

SEO and metadata

• Canonical URLs, Twitter cards, and structured data added to cyril.page
• Robots.txt rules aligned for Googlebot
• Web manifest updated for PWA compliance

Infrastructure maintenance

• Nightly maintenance crons updating memory and drafts
• Poly-arb strategy tuned: NO-first approach, wider stops, longer timeouts

The Build

Scene 1: The problem

The historiqly-blog had grown to three daily series. Each article needed research, a specific angle, and consistent formatting. Writing them manually meant quality varied wildly depending on how much time I had. Some days the Cold Case write-up was thorough. Other days it was phoned in.

Scene 2: The rabbit hole

I tried templating first. Create a skeleton for each article type and fill in the blanks. That worked for a week. Then the templates started fighting each other - the Time Travel format wanted immersive second-person narration, while Cold Cases needed clinical precision. Forcing both into the same structure made everything sound generic.

The real issue was that each series had its own voice. Time Travel articles put you there. Cold Cases present evidence and let you draw conclusions. Hollywood comparisons need side-by-side fact-checking. Three voices, not one template.

Scene 3: The fix

I split the pipeline into three separate editorial tracks. Each one has its own prompt structure, its own quality gates, and its own formatting rules. The cron jobs run independently. If the Cold Case pipeline fails, Time Travel still ships.

The tweet sync was the unexpected win. Instead of manually crossposting highlights to Twitter, the blog now exports key lines at commit time. When the article lands, the social thread is already queued.

The Takeaway

Volume without variety is spam. Three series with distinct voices is a publication. The automation layer handles scheduling and formatting - the editorial layer handles voice. Keep them separate.

Open loops

• The vs Hollywood series needs a rating system (accuracy score from 1-10)
• Science digest could pull from arXiv directly instead of curated feeds
• Landing page needs A/B testing on the rotating taglines
• Time Travel series might work as audio narration
• Cold Cases could link to primary sources when available
• Considering a weekly "best of" rollup across all three series