In a recent LinkedIn post, Ruben Hassid shares a comprehensive guide for users looking to enhance their proficiency with Anthropic’s Claude AI, emphasizing the importance of foundational knowledge and strategic prompting. Hassid, who taught his mother, a retired pharmacist, to use Claude, distills his experience into seven key lessons designed to help users move beyond basic interactions and unlock the AI’s full potential.
Understanding Claude’s Model Hierarchy
Hassid begins by highlighting a crucial point often overlooked by users: the existence of multiple Claude models. He stresses the importance of selecting the right model for the task at hand, cautioning against using the less capable Haiku 4.5 and Sonnet 5 models for complex operations, even if free. Hassid suggests starting with Fable 5 for initial prompts to define the task’s scope and desired outcome, then transitioning to the more powerful Opus-5 for execution.
“There is not one Claude, but four.”
As Ruben Hassid notes, understanding this model differentiation is fundamental. He further elaborates on the initial setup, recommending a specific prompt structure to guide Claude effectively:
“I need [task] for [goal]. I will know it worked when [target]. Ask me questions first, using the AskUserQuestion tool.”
Hassid emphasizes that users should leverage the clickable form Claude generates from this prompt and not shy away from providing ample context in the “something else” option, stating, “Don’t be lazy.”
Advanced Techniques and Avoiding Pitfalls
The post delves into more advanced strategies, including techniques to mitigate AI writing patterns and prevent Claude from becoming ‘dumber’ over time. Hassid introduces concepts like ‘Research mode’ for generating analyst reports, ‘Claude Artifacts’ for building shareable mini-applications, and a clever workaround for image generation using HTML to ensure text accuracy.
Leveraging Specialized Modes
Ruben Hassid points out several underutilized features. He highlights ‘Incognito mode’ as essential for generating genuinely new ideas without the influence of past interactions and recommends using the ‘Interactive chart’ feature after uploading a CSV file for data visualization.
“Claude is not built to make images. But if you ask it to make an image in HTML format. It’ll never get the text wrong on the images.”
A significant piece of advice from Hassid concerns prompt limits. He advocates for starting a new chat after approximately 40 prompts, explaining that Claude re-reads the entire chat history with each turn, which can lead to hallucinations and diminished performance. He suggests using the “Restart the conversation from here” option on a previous message to maintain efficiency.
Strategic Prompting and Usage Habits
Hassid also outlines specific behaviors to avoid, such as using Fable 5 for tasks better suited to Opus 5, which he warns can waste tokens. He advises using ‘Projects’ for repetitive tasks, new chats for creative work, and Incognito mode for context-free sessions. He also recommends converting long, frequently used prompts into Claude Skills.
Furthermore, Hassid cautions against prompting with excessive adjectives and instead suggests providing examples of poor writing and instructing Claude to avoid such styles. He advocates for sticking with what works rather than constantly chasing the latest features.
Optimizing Claude’s Rolling Window
Finally, Ruben Hassid touches upon Claude’s usage limits by explaining its rolling 5-hour window. He recommends strategically spacing out interactions throughout the day, suggesting about three sessions—one upon arrival at the office, and then one in the early and late afternoon—to maximize effectiveness.
Hassid concludes by offering a Claude skill library as a resource for those seeking deeper engagement with the AI, available to new free subscribers at how-to-ai.guide.
📝 About This Content
This article is based on insights shared by Ruben Hassid on LinkedIn.
📅 Originally posted on August 17, 2026 | View original post on LinkedIn →