Key takeaways
- The post you're reading right now was produced by the same pipeline it describes.
- Every morning it checks a YouTube feed, decides what's actually worth writing about, and drafts a real blog post from the transcript, not a copy-paste job.
- Most of what comes through the feed gets skipped on purpose. A curated pile beats a big pile every time.
- Nothing goes live without me. Every draft sits in GHL until I publish it, and every promo waits in an approval inbox until I say yes.
- The system tells on itself when it breaks, which is the only way I trust it to run unattended.
Here's a strange one. This post exists because a piece of software woke up this morning, checked a video feed, decided something in it was worth writing about, and wrote this. I didn't touch it until I sat down to read the draft.
That's not a gimmick. It's the actual daily job running on On Point Tech's operating system, and since I keep writing build logs about the systems we run here, it felt dishonest to skip the one responsible for the build logs themselves.
What it checks every morning
Every day the pipeline pulls the latest videos from a YouTube channel's public feed. That feed caps out at the fifteen most recent uploads, which is a real limitation worth knowing if you ever try to build something like this yourself, but for a daily check it's plenty. New video shows up, the pipeline looks at it. No new video, it says so and stops.
The feed isn't picky about what gets uploaded. Some days it's a real workshop or a masterclass. Other days it's a fifteen-second clip about an app update nobody outside the product team will ever care about. Blogging everything that comes through would be SEO spam, and Google can tell the difference between a business publishing real content and a business publishing noise for the sake of publishing something.
The editorial call is the actual work
So there's a filter. Masterclasses, real tutorials, spotlight interviews with actual substance, anything strategic or technical get written up. Release-note clips, app version bumps, event promos, and anything under two minutes get logged as seen and skipped. That filter is the part I actually care about getting right, because a blog full of thin posts hurts more than a blog with fewer, denser ones.
When something clears the bar, the pipeline pulls the full transcript and writes an original post from it. Not a transcript dump with paragraph breaks added. A real rewrite: key takeaways up top, a few sections that actually explain what mattered, a local angle for Bay Area and San Jose readers, practical steps a reader can act on, and the video embedded at the bottom for anyone who wants the full thing.
Nothing ships without a human
Here's the part that matters more than the writing. Every blog post lands as a draft, never published automatically. I'm the one who reads it and hits publish. And any social promo written to go with it doesn't touch our social accounts either, it sits in a local approval file until I say yes.
That rule exists because I learned it the hard way on a different part of this system. Early on, a batch of social posts got staged with a status that I assumed was a safe holding pattern, and the platform published them anyway before I ever looked at them. Nobody got hurt, but it was a real lesson: don't trust a status field to be a gate. Build an actual gate, in a file only I control, and check it by hand until it earns trust.
Why this matters for Bay Area and San Jose businesses
Most small businesses I talk to know they should be publishing content regularly and just don't, because "regularly" is the part that breaks. Someone gets busy, the blog goes quiet for six weeks, and Google notices the same way readers do. A system like this doesn't require more willpower than you have on your best day. It requires a source of raw material, a filter for what's actually worth publishing, and a human checkpoint before anything goes out. That pattern works whether the raw material is your own YouTube channel, your team's meeting notes, or your Google reviews.
Practical steps
- Pick one source of raw material you already generate that you're not reusing yet. Videos, call recordings, meeting notes, all of it is raw material for content.
- Write down your actual filter for what's worth publishing before you automate anything. If you can't describe the filter in a sentence, the automation won't know it either.
- Keep a human checkpoint before anything goes external. Automate the drafting, not the decision to publish.
- Expect it to surprise you at least once. Build the system to fail loud, not quiet, so you catch it fast.
What's coming next
This same pipeline just picked up a second and third lane: one writing reputation-management posts for our review card shop, and one writing these build logs. Three posts a day now come out of a system that started as a single YouTube-to-blog script three weeks ago.
We build systems like this at On Point Tech Solutions in Los Gatos, serving San Jose and the wider Bay Area. Every one of them, you own outright, done for you or taught with a full SOP so your team can run it without us. If your content is a "someday" project instead of a system, let's fix that. Start at optechsol.llc.