Maximize Claude’s $20 Plan: Ruben Hassid’s 21 Hacks for Cost-Effective AI Use

In a recent LinkedIn post, Ruben Hassid explores strategies for effectively utilizing the $20 per month Claude AI plan, cautioning users against immediately upgrading to the more expensive $100 tier. Hassid presents a detailed list of 21 “hacks” designed to optimize performance and token usage, making the lower-cost plan sufficient for most users’ needs.

Optimizing Input and Prompting Strategies

A key area Hassid addresses is how users interact with Claude, particularly concerning input size and prompt construction. He points out that uploading large PDF files directly consumes significant tokens. Hassid’s solution involves a more efficient workflow: pasting text into a Google Doc, downloading it in Markdown format, which drastically reduces the token count per page.

Furthermore, Hassid emphasizes the importance of concise prompting. He criticizes the common practice of writing lengthy prompts, suggesting a shift towards shorter, more directive instructions. As Ruben Hassid notes:

You write 500-word prompts that reload. Fix: Write 29 words instead: “I want to [task] to [goal]. Ask me questions using AskUserQuestion.”

This advice encourages users to frame their requests clearly and efficiently, prompting Claude to ask clarifying questions rather than attempting to process overly broad initial instructions.

Efficient Chat Management and Workflow

Ruben Hassid also delves into managing chat sessions to conserve tokens and maintain context. He advises against building files within Claude’s “Cowork” feature too early in the process, recommending planning in the chat interface first before moving to Cowork when the objective is clear. This prevents unnecessary token expenditure on preliminary planning.

Hassid advocates for breaking down complex tasks into smaller, manageable messages rather than sending multiple separate requests. According to Ruben Hassid:

You send 3 separate messages for 3 tasks. Fix: One message, three tasks. “Summarize this, list the points, suggest a headline.”

He also highlights the utility of the ‘Edit’ function on previous messages to correct errors or refine instructions, rather than resending entire prompts or issuing broad “redo” commands. This preserves context and avoids redundant processing.

Leveraging Claude’s Features and Understanding Limitations

The post further details how to best utilize Claude’s specific features and understand its limitations. Ruben Hassid suggests creating a prompt library to avoid rewriting prompts from scratch for recurring tasks, recommending a standardized structure with swappable variables. For deep work and complex tasks, Hassid differentiates between tools like Sonnet for quick checks and Fable for more intensive efforts.

Hassid addresses the issue of large “about-me” files by recommending they be trimmed and converted into Claude Skills. He also stresses the importance of restarting conversations when they go off track and periodically summarizing long chats to manage context effectively. According to Ruben Hassid:

New topic = new chat. Always. Dead context is dead tokens.

He also advises users to disable features like search and connectors by default, enabling them only when necessary for a specific task to manage token usage. For recurring tasks, Hassid points to Claude’s scheduling capabilities using commands like `/schedule`.

Strategic Tool Selection

Finally, Ruben Hassid underscores the importance of knowing the capabilities of different AI tools. He advises using Claude for its strengths in text generation and analysis, while directing image-related tasks to tools like ChatGPT and real-time search needs to platforms like Grok. This strategic selection ensures users are employing the right tool for the job, further optimizing efficiency and cost.