The company's database of screen-recorded webinars may be on the way.
Laura Tacho, CTO of developer productivity platform DX, says companies should at least consider rewriting documents to make them easier to read for AI.
Corporate documentation is often a nasty haze of notes, slide decks, and video training. In the age of AI, Tacho tells Business Insider that training materials should be in textual format, making them easy to read on large language models.
Text-based procedures are even more important for developers who can send documents to AI code editors such as Cursor immediately.
Tacho said he thinks he can do a document. It benefits both humans and LLMs. Talk about the topic of “practical engineers” Podcast, girlfriend The “AI-First” company said it is overhauling internal documents for LLM consumption.
“This was the biggest way businesses were thinking about services or were already trying to change the way they created services,” Tacho told podcast host Gergely Orosz.
In a follow-up interview with BI, Tacho explained her policy in detail for the documentary. Well-organized, text-based procedures don't just help LLM and human workers, she said. “The Venn diagram is a circle,” she said.
“The document is a major friction point for almost every organization,” Tacho told BI. “If a document is suitable for your purpose, there are many efficient benefits. This is a good area for LLM, which is good for humans.”
According to Tacho, human first documents often rely on “visual cues” and screenshots. She recommended that all photos have textual descriptions along with them.
Social media sites often have alternative texts and captions to make the service less visually visible. Tacho said corporate document makers can “take notes from the world of web accessibility.”
Related Stories
Business Insider tells the innovative stories you want to know
Business Insider tells the innovative stories you want to know
“It's very important for people who don't use their eyes to access the screen,” she said.
DX CTO Laura Tacho spoke with BI about what LLMS can read about the importance of ensuring internal documentation. Laura Tacho
She said other tasks, such as centralizing documents, help to make LLM easier access and also help to make life easier for human workers. Tacho used developer terminology and said companies should “cut down” their policies.
“The document is fragmented and I'm here a bit, a bit there,” she said. “You have to hop between different pages to be completely instructed on what you should do.”
Tacho also listed technical elements of the document, including verifying that HTML markup is semantic. This helps ensure readability of LLM.
Some companies have already begun making changes. In May, Vercel's Vice President of Developer Experience, Lee Robinson posted about the company's documentation on X.
“Anywhere you used to say “click” I've started adding Curl commands to Vercel documentation,” writes Robinson. “In the future, the computers using the agent may be able to log in and perform actions, which feels like a great incremental step for LLMS.”
Tacho hopes that more companies will join Vercel and make documents easier to read for AI. She discovered that engineers were wasted more than 30 minutes a week looking for information they couldn't find in the documentation. The AI code editor should automate these issues if LLM is able to properly read the document.
“It's actually a very important business issue when you think about how much time is wasted due to inadequate documentation,” she said.
