Communication Best Practices with LLMs

How to Work Effectively with AI Assistants

The Golden Rules

1. Be Specific, Not General Avoid vague requests like "Make it better." Instead, say "Make the text larger and add more spacing between paragraphs." Specificity leads to better results every time.

2. Give Context Instead of saying "Add a button," provide full context: "Add a 'Call Now' button in the top right corner that's green and stands out." Context helps the LLM understand your intent.

3. Reference Your Task List Rather than asking "What should we do next?" try: "We just completed task 3. Looking at our list, task 4 is 'Create menu page'. Are we ready for that or should we break it down further?"

4. Ask for Alternatives Say things like "Can you give me 3 different ways to display the menu items?" or "What are the pros and cons of each approach?" This ensures you explore the best options.

5. Set Expectations Let the LLM know your constraints: "I'm a beginner, so please explain technical terms," "I need this to work on mobile phones too," or "I have a budget of $0, so suggest free solutions."

The Check and Confirm Pattern

Before big changes, always follow this pattern: - Summarize: "So we're going to add a contact form with name, email, and message fields" - Confirm: "Is this what you want, or should we change anything?" - Proceed: "Great, let's start with task 1: designing the form layout"

When You Get Stuck

Use these phrases: - "I don't understand this part. Can you explain it differently?" - "This seems too complicated. Can we try a simpler approach?" - "Can you show me what this would look like to a visitor?" - "What's the easiest way to do this?"

Advanced Communication Techniques

The "Show Me" Method: Instead of just describing, ask the LLM to show examples. Request different layouts, professional-looking forms, or how other businesses display their content.

The "Step-by-Step" Method: For complex tasks, ask for detailed breakdowns. Say "Can you break down 'add a contact form' into specific steps?" or "Walk me through setting up online payments step by step."

The "What If" Method: Explore alternatives and edge cases with questions like "What if a customer can't find our phone number?" or "What happens if someone submits the form without an email address?"

Effective Feedback Techniques

When something is not right, do not struggle in silence. Describe the problem specifically, reference your task, and ask for alternatives. Give constructive feedback like "The homepage looks great, but the menu section is hard to read. Can we make the text larger and add more contrast?"

Use iterative improvement: "This is good, now let's make these 3 small improvements..." or "I like the layout, can we try a different color scheme?"

Working with Multiple LLMs

Different LLMs have different strengths. You can use ChatGPT for brainstorming and content creation, Claude for detailed analysis and planning, and Cursor for actual code implementation. Always share your task list between different LLMs so they understand your progress.

Key Takeaways

• Be specific -- vague requests produce vague results
• Provide context -- help the LLM understand your project
• Reference tasks -- keep your task list central to conversations
• Ask for alternatives -- do not accept the first suggestion
• Give feedback -- be specific about what works and what does not
• Iterate -- build in small, refinable steps

Effective communication is the foundation of successful LLM collaboration. Master these patterns, and watch your productivity soar.