In a recent LinkedIn post, Hiten Shah discusses a fundamental shift in how we should think about the outputs of Artificial Intelligence, arguing that the current text-centric approach is insufficient for many practical applications. Shah highlights that AI, while adept at generating text-based content, often stops short of creating truly usable formats for end-users.
According to Shah, the initial phase of AI’s utility has been in generating summaries, plans, outlines, tables, and drafts. While valuable, this is only the first step. He points out the limitations of this text-first approach:
“Text is easy to generate, copy, edit, and judge. Then the real work starts. Someone has to find the signal, rebuild the structure, and turn the answer into something another person can use.”
Shah elaborates on the shortcomings of traditional document formats when dealing with complex information, stating that many documents fall short because they function as weak interfaces. He provides examples of where text-based outputs often fall flat:
“Research reports benefit from retrieval. Pricing memos rely on adjustable assumptions. Launch plans require coordination. Product ideas call for mockups. Strategies depend on a clear path through the logic.”
This is where Shah sees a resurgence in the importance of HTML. He posits that the web browser, a ubiquitous tool already installed on most devices, is the ideal environment for AI-generated content to become truly actionable. Shah argues that AI can now leverage HTML to create self-contained, interactive pages.
The Interactive Potential of HTML for AI
Hiten Shah emphasizes that AI’s ability to generate interactive HTML pages represents a significant evolution beyond static text documents. This new format allows users to engage directly with the information in ways that are not possible with traditional text outputs. As Hiten Shah notes, these pages offer a more dynamic and functional experience:
“AI can now turn source material into a self-contained page someone can open, scan, click through, adjust, and share.”
This interactive capability is crucial for various business functions. For instance, instead of a static pricing table, an AI could generate an HTML page with adjustable assumptions that a user can manipulate. Similarly, product ideas could be presented not just as text descriptions but as interactive mockups within an HTML page, providing a much clearer and more engaging user experience.
Shifting the AI Prompting Paradigm
Shah encourages a fundamental change in how users interact with AI tools. Instead of simply asking for information, he suggests focusing on the intended use and the end-user’s needs. This user-centric approach to prompting, he argues, will unlock the full potential of AI.
As Hiten Shah puts it:
“Before asking AI for another doc, ask what the output needs to help someone do.”
This perspective reframes AI from a mere content generator to a tool capable of creating functional, interactive applications. By focusing on the desired action or outcome, users can guide AI to produce outputs that are not just informative but also directly useful and adaptable, moving beyond the limitations of plain text and embracing the interactive possibilities of the web.
📝 About This Content
This article is based on insights shared by Hiten Shah on LinkedIn.
📅 Originally posted on July 9, 2026 | View original post on LinkedIn →