In a recent LinkedIn post, Ruben Hassid shares a comprehensive set of strategies for getting the most out of AI assistants like Claude, drawing from extensive personal use. Hassid, who has logged over 1,800 hours interacting with Claude, frames his advice as essential ‘reps’ that users shouldn’t skip, akin to a system used by Marines for critical tasks.
Hassid emphasizes proactive engagement and specific prompting techniques to enhance the AI’s utility. One of his key recommendations involves making the AI ask clarifying questions before providing answers.
“Before answering, use the AskUserQuestion form to get more context from me if necessary.”
As Hassid notes, this approach, detailed as tip #8 in his compilation, ensures that the AI has a more complete understanding of the user’s needs, thereby reducing the chances of a misaligned or unhelpful response. He calls this the “most underrated feature.”
Transforming Interaction with AI Assistants
Beyond structured prompting, Hassid advocates for a more natural interaction style. He suggests using voice input through the Wispr Flow app, allowing users to speak their thoughts, constraints, and even contradictions for up to 10 minutes straight. This method, according to Hassid, helps preserve conversational context, which can be inadvertently lost when relying solely on typing and deleting.
Leveraging AI for Content Creation and App Development
Hassid also reveals a creative method for generating visual content without traditional AI image generators. By uploading an existing image to Claude and prompting it to “Code an HTML-like infographic like the one I attached, but about [topic],” users can then export the HTML to design tools like Canva.
“The text is always right, which image generators still can’t promise.”
This technique, highlighted as tip #3, ensures textual accuracy, a common challenge with current image generation models. Furthermore, Hassid demonstrates how to build rudimentary applications within a single message, integrating AI coaching and data persistence.
Optimizing Claude’s Performance and Settings
A significant portion of Hassid’s advice focuses on optimizing the AI’s performance through specific settings and usage patterns. He details how to manage connectors, adjust chat turn limits, and correctly edit previous prompts to rectify errors.
“When Claude is wrong, don’t type ‘no, that’s wrong.’ Go back to the prompt before the bad answer, edit it, save.”
According to Hassid, failing to edit the preceding prompt means the incorrect information remains embedded in the conversation’s history, potentially influencing future outputs. He also cautions against using large ‘Projects’ for novel idea generation, suggesting that Claude may default to the provided files rather than engaging in fresh, contextual thinking. Hassid posits that new ideas are best explored in empty, fresh chats.
Refining Prompts Through Negative Constraints
Hassid also offers a counter-intuitive but effective method for refining AI output: focusing on what to avoid rather than what to achieve. Instead of vague instructions like “Make it punchier,” he recommends providing examples of undesirable text.
“Never write like this: [paste the thing you hate].”
In Ruben Hassid’s view, this direct negative constraint clearly defines the boundaries for the AI, leading to more precise and satisfactory results than generic positive instructions.
These nine tips, representing a fraction of his full guide, underscore Hassid’s belief in a more deliberate and informed approach to interacting with advanced AI tools, transforming them from simple response generators into powerful, interactive collaborators.
📝 About This Content
This article is based on insights shared by Ruben Hassid on LinkedIn.
📅 Originally posted on July 30, 2026 | View original post on LinkedIn →