Documentation deserves its AI reckoning
Nobody wants to admit it, but most documentation is disposable.
User guides, help systems, knowledge base articles — they’ve been rotting for years. Bloated PDFs nobody reads. Auto-generated HTML help stuffed with screenshots from a version ago. Knowledge bases that feel like corporate junk drawers, where the “search” function is just a punishment for asking a question.
Generative AI is coming for that work — and it should.
Let’s be honest: the average doc set is a museum of absurdity. “Known issues” sections that haven’t been updated since the beta. Release notes that read like a Kafka short story. Step-by-step tutorials written for a UI that no longer exists. Half the time, the answer is buried in a Jira comment that never made it into the official help.
And how do those docs get built? An engineer dumps bullet points into Confluence. A tech writer converts them into sentences. A manager insists on screenshots for every step, even though the UI team is already redesigning half the product. By the time it ships, the doc is outdated. By the next release, it’s a liability.
The real problem? Documentation was always about compliance theater, not user empowerment. A checkbox for “customer enablement.” A vanity metric about page counts. A way for execs to say, “See, we support our users” while forcing them to click through a labyrinth of stale articles. Customers never wanted a 60-page PDF. They wanted an answer, right when they hit a wall.
That’s exactly where generative AI shines. Models can already draft contextual answers, update KB articles in real-time, and localize text instantly. No human writer should waste hours updating screenshots for a feature flag that may be removed in the next sprint. AI can bulldoze the busywork, and frankly, it should.
Here’s what nobody’s admitting: traditional documentation is the perfect low-hanging fruit for AI. The filler. The “update later” sections. The ritualized bloat nobody will miss.
Docs were supposed to empower users. Instead, they became cemeteries of outdated knowledge. Generative AI isn’t killing documentation — it’s writing the headstone.
