Research and Prompting

Research and Prompting

This guide shows how to use WILsquareAI for research. It focuses on clear tips, prompt design, and how to work with four models: o3, GPT-4o, Claude 3.7, and Gemini.

Start Wide, Then Focus

  • Begin with GPT-4o if you need an overview or initial summary.
  • After you have a foundation, switch to a more specialized model if your topic is complex or you have a large amount of text.
  • This approach helps you see the general picture before diving into specifics.

Get Live Data When Needed

  • If you need current facts, use a model with web access, such as Perplexity.
  • Ask for citations or links if you want to confirm sources.

Use Gemini for Huge Files

  • Gemini can handle larger texts or datasets in one go.
  • Let the model know what kind of content it is processing. It might be long reports, contracts, or data tables.

Complex Logic and Reasoning

  • Claude 3.7 and o3 can address more difficult topics or multi-step problems.
  • Request a detailed breakdown if you want a step by step explanation.

Compare Answers from More Than One Model

  • Asking the same question to different models can reveal helpful insights or details you might have missed.
  • Use what you like from each response to strengthen your final outcome.

Writing Good Prompts

  1. Be clear about the goal. For example, you might say, "Summarize this document with a focus on financial metrics."
  2. Include any context you have. If you have notes or key points, place them at the start of your request.
  3. Say how you want the response formatted. Bullet points, a concise list, or short paragraphs can be helpful.
  4. Ask follow-up questions. If you need more detail, keep refining your prompt.

Combining Models in Clever Ways

  • You can start with Gemini to review a large amount of text, then shift to o3 for real-time facts or additional verification.
  • If you have financial or numeric data, try GPT-4o or Claude 3.7 for a deeper look, then confirm sources with o3 if needed.

Quick Reference

  • Broad Research or quick summaries: GPT-4o
  • Live Data: Perplexity
  • Very Large Files: Gemini
  • Complex Logic: Claude 3.7 or o3
  • Multiple Angles: Compare answers from two or more models

Adjust your prompts and questions as you learn more about your topic. By staying specific in your instructions and clear about your goals, you can get useful research results from WILsquareAI.

Basic Chat and ModelsAdvanced Research Tools