In a recent LinkedIn post, Ruben Hassid offers a counterintuitive strategy for optimizing interactions with AI models like Claude, advocating for longer, more comprehensive initial prompts to save on computational ‘tokens’ and improve output quality. Hassid challenges the common practice of using short, iterative prompts, arguing that this approach often leads to token waste through AI re-reads and subsequent corrections.
Hassid’s core argument is that AI models, including Claude, re-evaluate the entire conversation history with each new message. This means that even a simple follow-up prompt requires the AI to process all previous exchanges, consuming tokens. He suggests that this is where the inefficiency lies, particularly when short prompts lead to misinterpretations and require further clarification.
“Stop writing short prompts to save Claude tokens. Write the longest prompt you can.”
The ‘Short Prompt Trap’ vs. The ‘Long Prompt Fix’
Ruben Hassid identifies a common pitfall he terms the ‘short prompt trap.’ This occurs when users employ vague instructions like “make it better,” forcing the AI to guess the desired improvements. According to Hassid, this guessing game often results in incorrect outputs, necessitating additional prompts and context, which in turn consumes more tokens.
Hassid contrasts this with his recommended ‘long prompt fix.’ This method involves providing all necessary information—the task, context, desired format, and tone—in a single, detailed initial prompt. He suggests using voice-to-text tools and keyboard shortcuts to expedite the creation of these comprehensive prompts, thereby increasing speed and reducing the need for back-and-forth communication.
Leveraging Voice and Speed for Efficiency
A key element of Hassid’s strategy is the utilization of tools that bypass traditional typing. He recommends installing applications like Wispr.ai, which allow users to speak their entire prompt in one go. By setting a custom shortcut key, users can activate this feature seamlessly within their AI interface.
“You’re not lazy because you write short prompts. You write short prompts because typing is slow. So stop typing.”
As Hassid explains, this approach ensures the AI has all the required context upfront. This allows the AI to generate a more accurate and relevant response on the first try, minimizing the need for follow-up messages that trigger the costly re-reading of conversational history.
The Economic Case for Comprehensive Prompts
Hassid frames token consumption as a direct monetary cost. He argues that frequent re-prompts and corrections, often stemming from initial brevity and vagueness, are the true drivers of expense when interacting with AI.
“A short prompt = a wrong guess = ‘no, I meant…’ = a reload. Every reload is tokens. Tokens are money.”
By investing time in crafting a single, detailed prompt, users can potentially save money and time in the long run. Hassid points out that a well-structured, long prompt allows the AI to move directly to generating the desired output, rather than engaging in a cycle of clarification and revision. He illustrates this with an example of a specific follow-up instruction:
“Tone’s too stiff. Make it sound like I’m texting a friend who runs a 200-person company. Keep the data. Only redo section 2.”
In Ruben Hassid’s view, this method transforms a potentially lengthy series of interactions into a single, efficient command. By embracing longer prompts and leveraging tools that facilitate faster input, users can reportedly achieve better results and manage their AI resource consumption more effectively.
📝 About This Content
This article is based on insights shared by Ruben Hassid on LinkedIn.
📅 Originally posted on June 17, 2026 | View original post on LinkedIn →