Laura Tacho, CTO of developer productivity platform DX, believes that corporate documentation should be understandable to both humans and AI. She pointed out that LLMs are not yet very good at distinguishing between screenshots and visuals, and recommends adding text descriptions to them.
Laura Tacho, CTO of developer productivity platform DX, believes that corporate documentation should be understandable to both humans and AI. She pointed out that LLMs are not yet very good at distinguishing between screenshots and visuals, and recommends adding text descriptions to them.
Tacho told the Pragmatic Engineer podcast that companies should at least consider rewriting their documents to make them more understandable to AI. In the age of AI, she said, training materials should be in text form so they can be easily read by a large language model.
As Business Insider notes, text formats are even more important for developers, who can immediately transfer their documentation to an AI code editor like Cursor.
Laura Tacho believes that documents can be useful for both humans and large language models. She noted that companies that put AI first are reviewing their internal documentation for use by large language models.
«Documentation is a huge source of conflict for almost every organization. When documentation is better suited to its purpose, significant efficiency gains can be achieved. This is an area where what’s good for the person is also good for the LLM,» Tacho told Business Insider.
Human-only documentation often relies on «visual cues» and screenshots that AI may find confusing. That’s why Laura Tacho recommends adding text descriptions to all images.
Recall that recently, a Swiss IT company that develops privacy-focused services, including Proton Mail, introduced an artificial intelligence chatbot called Lumo. It can summarize documents, generate code, write emails, etc., while storing data locally on users' devices.
