DX CTO Says Companies Should Rethink Corporate Documentation, Write Things AI Can Easily Read

The era of poorly organized memos, scattered slide decks, and screen-recorded webinars could soon be over. Companies are being urged to overhaul their internal documentation to make it legible for artificial intelligence systems, with developer productivity experts like Laura Tacho, CTO of the platform DX, sounding the warning on what she calls a hidden efficiency crisis.

Speaking with Business Insider and on The Pragmatic Engineer podcast hosted by Gergely Orosz, Tacho said many firms are already shifting toward what she described as “AI-first” documentation practices—writing and structuring company knowledge so that large language models (LLMs) can easily consume and act on it.

“Documentation is a huge point of friction for nearly every organization,” Tacho told BI. “There’s a lot of efficiency gain to be had when documentation is more fit for purpose. This is an area where what’s good for the human is also good for the LLM.”

According to Tacho, engineers lose over 30 minutes a week searching for information buried in cluttered or inaccessible documents—a problem that LLMs could solve if they were simply given structured text to read. But video trainings, screenshots, and visually reliant content block AI tools from accessing essential knowledge.

From Human-First to AI-Literate

Human-friendly documentation often leans heavily on visual design, relying on screenshots, diagrams, and interfaces to convey information. But these visuals, Tacho warns, are mostly illegible to AI. She recommends that every image be paired with textual descriptions, borrowing a principle from web accessibility standards, such as alternative text and captioning used by social platforms for the visually impaired.

“It is so important for people who are not using their eyes to access the screen,” she said, emphasizing that practices that improve accessibility also unlock AI comprehension.

The concept, she says, is not just about helping LLMs. “The venn diagram is a circle,” Tacho noted, highlighting that clear, structured documentation improves productivity for both humans and machines.

Beyond that, companies are being advised to centralize their documentation, avoiding the fragmented sprawl that forces users to navigate multiple internal portals to complete a single task.

“Documentation is made piecemeal, a little here, a little there,” she said. “You have to hop between different pages in order to put together the complete instruction of what you’re supposed to do.”

The solution? Tacho calls it a “defrag” of internal policies—borrowing the term from computing to suggest restructuring for efficiency.

Code-Ready Language and Real-World Examples

For software companies in particular, making documentation LLM-compatible could soon be non-negotiable. Developers are already plugging documentation directly into AI code editors like Cursor, which allow them to generate code snippets and explanations instantly. The catch: the documentation has to be in readable, structured text.

Some companies are already adapting. In May, Lee Robinson, former VP of Developer Experience at Vercel, revealed changes to the company’s documentation strategy on X (formerly Twitter).

“We’re starting to add cURL commands to Vercel’s documentation wherever we previously said ‘click,’” he wrote.

Robinson suggested that in the near future, AI agents might not just read documentation, but act on it—performing tasks, logging in, and executing commands on behalf of users.

That’s a future Tacho also envisions—but it won’t happen unless businesses start writing for both people and machines. She cited technical standards like using semantic HTML markup to help LLMs better parse and understand online documentation.

A New Corporate Priority

What might sound like a marginal improvement in productivity could have significant implications for company performance. With LLM tools like OpenAI’s ChatGPT or Anthropic’s Claude becoming common in enterprise workflows, the quality and accessibility of internal documentation could directly impact how much return companies see from their AI investments.

“When you think about how much time is being wasted due to poor documentation, it becomes actually a very critical business problem,” Tacho said.

As AI systems continue to evolve, companies still stuck with scattered training videos and PDF slide decks may find themselves not just inefficient, but incompatible with the next wave of workplace automation. The age of screen-recorded onboarding is waning, and text, it seems, is the new code.