How to Write Better AI Prompts: 10 Tips

Getting great results from AI models is not luck — it is technique. The difference between a mediocre output and a brilliant one often comes down to how you write your prompt. After testing thousands of prompts across ChatGPT, Gemini, Claude, and ChatLLM DeepAgent, we have distilled our best advice into 10 actionable tips.

Tip 1: Be Specific About What You Want

The number one mistake people make is being too vague. "Write a blog post" gives the AI almost nothing to work with. Instead, try: "Write a 1,000-word blog post about the top 5 productivity apps for remote workers in 2026. Use a professional but friendly tone. Include an introduction, 5 subheadings (one per app), and a conclusion."

The more details you provide, the closer the output will be to what you actually need. Specify the topic, length, tone, structure, audience, and any constraints.

Tip 2: Assign a Role or Persona

AI models perform dramatically better when you give them a role. "You are a senior copywriter at a Fortune 500 tech company" produces different (and usually better) output than a generic request.

Some powerful role assignments: - "You are an expert SEO content strategist." - "You are a data scientist explaining concepts to a non-technical audience." - "You are a startup mentor with experience in SaaS businesses."

The role sets the context, expertise level, and tone for the entire conversation.

Tip 3: Use Chain-of-Thought Prompting

For complex tasks, ask the AI to think step by step before giving its final answer. This technique, called chain-of-thought prompting, dramatically improves accuracy on reasoning tasks.

Instead of: "What is the best marketing strategy for a new SaaS product?"

Try: "I am launching a new SaaS product for project management. Think step by step about the target audience, competitive landscape, budget constraints of a startup, and then recommend a marketing strategy with specific channels and tactics."

Tip 4: Provide Examples (Few-Shot)

One of the most effective prompting techniques is providing examples of what you want. This is called few-shot prompting.

For instance, if you want product descriptions in a specific style, provide two or three examples first, then ask the AI to write new ones in the same format. This teaches the model your preferred style, structure, and tone far more effectively than describing it in words.

Tip 5: Specify the Output Format

Do not leave the format to chance. Tell the AI exactly how you want the response structured:

• "Respond in a numbered list."
• "Format your answer as a markdown table with columns for Name, Price, and Rating."
• "Return the result as valid JSON with keys: title, description, tags."
• "Write this as a bulleted outline, not full paragraphs."

Specifying format is especially important when you are feeding AI output into another tool or workflow.

Tip 6: Set Constraints and Boundaries

Constraints help the AI focus. Without them, responses tend to be generic and overly broad.

Effective constraints include: - Word or character limits: "Keep your response under 300 words." - Time boundaries: "Only include information from 2025 and 2026." - Scope limits: "Focus only on free tools. Do not mention paid options." - Exclusions: "Do not include any disclaimers or caveats."

Constraints force the model to prioritize and make decisions, which usually produces more useful output.

Tip 7: Iterate and Refine

Professional prompt engineers rarely get the perfect output on the first try. Iteration is a core part of the process.

Start with your best prompt, review the output, and then refine. You might say: "This is good, but make the tone more casual" or "Add specific metrics and data points to each section" or "Reorganize this with the most important points first."

Each iteration gives the AI more information about your preferences. Think of it as a conversation, not a one-shot request.

Tip 8: Use System Instructions When Available

Many AI platforms — including ChatGPT, Google AI Studio, and Claude — support system instructions. These are persistent rules that apply to the entire conversation.

Use system instructions for: - Setting a permanent role or persona - Defining output format rules - Establishing tone and style guidelines - Specifying what to avoid

System instructions are like giving the AI a job description before it starts working.

Tip 9: Break Complex Tasks Into Steps

If you need a complex output — like a full marketing plan, a business proposal, or a multi-section report — break it into individual prompts rather than asking for everything at once.

Step 1: "Research the target audience for a meal prep delivery service aimed at busy professionals." Step 2: "Based on that audience analysis, suggest 5 marketing channels with pros and cons." Step 3: "Create a 30-day social media content calendar using the top 3 channels."

Each step builds on the previous one, and you can course-correct along the way.

Tip 10: Learn the Strengths of Each Model

Different AI models have different strengths. Matching your task to the right model is part of effective prompt engineering.

• ChatGPT (GPT-5): Excellent for creative writing, brainstorming, and conversational tasks.
• Google Gemini 2.5 Pro: Best for long-context analysis, multimodal tasks, and structured output.
• Claude: Outstanding for careful reasoning, nuanced writing, and following complex instructions.
• ChatLLM DeepAgent: Ideal for multi-step workflows and building complete applications.

Putting It All Together

Here is an example of a prompt that uses multiple tips at once:

"You are a senior content strategist specializing in B2B SaaS marketing (Tip 2). Write a 1,200-word blog post about lead generation strategies for early-stage startups (Tip 1). Structure the post with an introduction, 5 subheadings, and a conclusion (Tip 5). Use a professional but approachable tone. Only include strategies that cost less than $500 per month to implement (Tip 6). Think step by step about which strategies have the highest ROI before writing (Tip 3)."

That single prompt incorporates five of our ten tips and will produce a dramatically better result than "Write about lead generation."

Practice these techniques, and you will see an immediate improvement in the quality of AI outputs you receive. Browse our prompt library for ready-made templates that incorporate all of these best practices.