Why this exists
PlainQuill comes out of practice, not a pitch deck. The pipeline was built to solve a real problem on real sites, and it ran there long before it had a name.
A year of making AI writing hold up
PlainQuill’s founding engineer spent more than a year running a daily AI content pipeline on his own sites, and most of that time went into one problem: making the writing reliable enough to publish under a name he had to stand behind. That meant research from real sources with stored snapshots, automated checks that fail closed, a second model from a different vendor verifying the first, and a human approval step that nothing skips.[1] The checks were not a feature idea. They were repairs, added one at a time, each after something slipped through and had to be caught by hand. The traffic grew because the content held up.
The service was Claude’s idea
Selling it was never the plan. During a conversation about a career change, Claude, the AI assistant, pointed out that the pipeline itself was the product: plenty of tools generate articles, and almost none of them check those articles against sources before a customer’s name goes on them. It took another two months to warm up to the idea of offering it as a managed service. What changed the founder’s mind was the advertising. Ads for competing services kept arriving, and reading the vendors’ own pages showed what they charged and what they left out. One sells auto-publish as the default. None advertises sources on the page. Extras like better images cost more on top.[2] The pipeline already did the things those pages did not mention, so the case for selling it stopped being hypothetical. This site is the result of that decision.
What that history buys you
Honesty about the cost, first. Research and grounding are the expensive part of this pipeline. Fetching sources, snapshotting them, tracing every claim back to a passage that supports it and paying a second model to check the first all cost more than generating plausible text and hitting publish. Keeping that affordable at a flat monthly price took deliberate engineering, and it is why the service runs on efficient models disciplined by process rather than expensive models taken on faith.[3]
And second, a product whose defaults were set by someone who had to live with the output. Approval is mandatory because the operator wanted it for his own sites. Sources are on the page because readers deserve to check them. The checks are deterministic because a check that can quietly stop checking is worse than none.
Start the trial
Run the same pipeline on your site
A week of real output on your own topics costs $3. Read the drafts, open the sources, keep what you approve.