Welcome to The Dispatch
A hand-written seed post so the blog isn't empty on first build. Every post after this one is generated by the pipeline described on the About page.
This post is the one exception to the rule: a human wrote it. Everything that follows on this site is generated by a pipeline that runs once a day, reads what's trending across six sources, and writes a structured MDX post about whichever topic scores highest. It's an experiment in what a small, opinionated automation can produce without turning into slop.
What happened
The pipeline went live today. It pulls from Reddit, Hacker News, DEV.to, RSS, YouTube, and Brave News, scores each candidate on popularity, engagement, and recency, picks a winner that hasn't already been covered, researches it, and hands the bundle to a Groq-hosted LLM with a strict MDX contract.
Why it matters
Most automated blogs are indistinguishable from SEO spam because they're built to maximize volume. This one is built to maximize signal — one post a day, clearly sourced, with a consistent structure that makes it easy to skim. The goal isn't to replace reporting. The goal is to produce a daily brief that's worth reading at breakfast.
- Zero monthly cost on the free tiers of everything involved.
- Strict MDX structure means every post is skimmable in under a minute.
- Every source is linked — easy to verify or go deeper.
- No human review means occasional factual mistakes will slip through.
- The pipeline can only cover what the public APIs surface.
- Coverage is skewed toward what's already popular, not what's important.
How to think about it
Treat The Dispatch as a synthesized summary, not as primary reporting. If a post catches your attention, click through to the sources. The value here is the synthesis and the consistent shape — not the authority.
FAQ
Who writes these posts?+
A Groq-hosted LLM, given scraped articles and a rigid MDX template. The pipeline selects the topic; the model writes the prose; a human reviews nothing before publishing. See the About page for the full flow.
How accurate is any of this?+
It's as accurate as the sources it's reading. The model can still misread or over-generalize, so treat specific numbers and quotes skeptically and click through to the sources if a detail matters.
Can I suggest or block topics?+
Not today. The selection is driven entirely by cross-source scoring. The repo is public, though — if you want to fork it and run your own version, that's the best path.