Prompts

How Prompts Work

You don’t need special syntax or code. Think of a prompt as giving directions to an assistant. Useful prompts often include:

• Goal: What you want
  • “Summarize this document…”
  • “Help me write an email…”
  • “Explain this like I’m new to the topic…”
• Context: Background or details
  • What you’re working on
  • Who it’s for
  • Any constraints or preferences
• Format: How you want the answer
  • Bullet points, steps, a table, an email draft, etc.
  • “Keep it short,” “Use non-technical language,” or “Include examples”
• Tone: If it matters
  • Friendly, professional, formal, casual, etc.
  • Example of a simple prompt:
  • “Write a short, friendly email to a coworkers promoting our newest website.”

Improving Workflow

• Efficiency: Good prompts can help save time by quickly getting you the information you need without endlessly sifting through unrelated results. Instead of typing random queries, take a moment to think about how to phrase your request better.
• Task Management: You can use prompts to help with various tasks. For instance, if you’re looking for project management tools, you might ask, “What are the top project management tools for small teams?” This way, you can receive tailored suggestions that fit your needs.

If you’re not sure what to type, you can literally prompt: “I’m trying HopGPT for the first time. Ask me a few questions and then suggest some prompts I can use.” HopGPT will guide you from there!

Chain of thought prompting is a powerful technique that enhances AI reasoning by guiding the model through intermediate steps rather than jumping straight to a final answer.

What is of chain of thought?

Instead of asking for an immediate solution, chain of thought prompting breaks complex problems into sequential, logical steps. This structured approach helps the AI “think through” the problem methodically, leading to more accurate and reliable results.

Why use chain of thought?

When working with complex tasks, vague instructions often produce unclear results. Chain of Thought prompting provides:

• Improved accuracy through step-by-step reasoning
• Greater transparency in the AI’s decision-making process
• Easier troubleshooting when results need adjustment

Example: Dataset cleaning

Less effective approach: “Clean this dataset.”

Chain of thought approach:

• Step 1: “Check if all entries in the date column follow this format: YYYY-MM-DD”
• Step 2: “Check if all entries in the weight column are between 80 and 150”

Best practice

When tackling complex tasks with HopGPT, break them down into clear, sequential steps. Guide the model through your reasoning process, and you’ll see significantly better outcomes.

First, open the Prompts panel.

• The “Prompts” icon is located at the top left of HopGPT. Click this icon to open the “Prompts” panel. 
• When the panel opens you’ll see a list of saved prompts (if any). Each entry usually shows the prompt title and a short description. Use this list to choose an existing prompt or to confirm that the one you want to add is not already present.

Next, create a new prompt.

• Click the “+ Create Prompt” button.
• A prompt editor form appears with fields such as:
  • Prompt Name: a short, descriptive name for the prompt.
  • Text: the full instruction or query you want the model to use.
  • Description / Notes (optional): extra context about when to use this prompt.
  • Category (optional): e.g., “Learning”, “Code”, “Roleplay”.
• Enter a clear title and paste or type the prompt text. Keep instructions concise but specific. Be sure to include role, format, constraints, and examples if helpful.
• HopGPT supports placeholders, include them in the prompt text using the expected syntax (for example: {{topic}} or {input}). Explain in the description how to replace them when running the prompt.
• Click “Create Prompt”. The new prompt appears in the panel list immediately.

Now, test your prompt.

• Select the prompt from the list to open a run view or to load it into the main chat input.
• If the prompt has placeholders, fill them in the provided fields.
• Click “Submit” to run your prompt.
• If the response needs refinement, return to the prompt editor to tweak wording, add constraints, or include an example. Save a new version or update the existing prompt.
• For commonly used variants, duplicate a prompt (use “Duplicate” or “Save as new”) and modify the copy.

Ready-to-use prompts at your fingertips

Not sure where to start? HopGPT’s Prompt Library offers a curated collection of pre-written, tested prompts designed to help you get the most out of your AI interactions.

Why use the prompt library?

• Save time: No need to craft prompts from scratch
• Learn by example: See how effective prompts are structured
• Get better results faster: Use proven templates that work
• Discover new use cases: Explore prompts for tasks you hadn’t considered