Mastering AI Assistants: Ruben Hassid’s Guide to Effective Prompting

In a recent LinkedIn post, Ruben Hassid explores how users can move beyond basic interactions with advanced AI models like Claude to unlock their full potential. Hassid, a proponent of structured AI engagement, argues that many users treat these powerful tools like simple chatbots, failing to leverage their ‘godlike intelligence’ due to ineffective prompting strategies.

Hassid outlines a series of common pitfalls and provides actionable advice for users aiming to get more sophisticated results from AI assistants. He emphasizes that the quality of the output is directly tied to the quality of the input and the user’s approach.

“You ask before you aim: Claude can’t hit a target you never named. Start with the task. ‘I want [X] for [Y].'”

The Importance of Clear Objectives

A core theme in Hassid’s post is the necessity of defining clear objectives before engaging with the AI. He points out that AI cannot fulfill a request if the user’s goal is ambiguous. Hassid advises users to be explicit from the outset, stating precisely what they want to achieve and for whom. This initial clarity, according to Hassid, sets the foundation for a productive interaction.

Avoiding Assumptions and Encouraging Dialogue

Ruben Hassid also highlights the danger of letting the AI make assumptions. When users do not provide enough context or explicitly ask the AI to seek clarification, it will fill in the blanks with its own interpretations, which are often incorrect for the specific use case. Hassid suggests an alternative:

“You let it guess instead of ask: It fills gaps with assumptions. Bad ones. Say ‘but first, ask me questions.'”

This approach, Hassid argues, transforms the interaction from a one-way command into a collaborative dialogue, ensuring the AI’s responses are aligned with the user’s actual needs.

Iterative Refinement Over Rework

Furthermore, Hassid challenges the common tendency to accept the first output from an AI or to start the entire process over when the initial result is unsatisfactory. He advocates for an iterative process of refinement. Instead of discarding a subpar response, Hassid recommends asking for multiple versions and then providing specific feedback to guide the AI toward a better outcome.

“You accept the first answer: The first one is never the best one. Ask for 3 versions. Pick. Refine.”

He also advises against restarting conversations when an AI misses the mark, suggesting instead that users clearly articulate what was misunderstood and provide the correct direction. This targeted feedback is more efficient than starting a new thread.

Personalization and System Building

Ruben Hassid stresses that to make AI outputs sound authentic and personalized, users should provide the AI with their own information, such as an ‘About-Me’ file. This helps the AI adopt a specific voice and style, moving away from generic responses.

Ultimately, Hassid’s post is a call to action for users to develop systems for interacting with AI, rather than treating each session as an isolated event. He suggests that by saving templates and refining workflows, users can significantly enhance their productivity and expertise with AI tools. He concludes by offering a pathway for readers to access his templates and resources, encouraging them to become the ‘AI guy’ in their organizations.