Sheet 2

Which tool, which model

AI Tools Field Guide

Not affiliated with OpenAI, Google, xAI or Anthropic; your organization’s rules come first.

Pick the tool by what only it can do, then pick the thinking level.

Four tools, and what only each does

  • Gemini (Google). Deep Research across the open web, a large context window, and Gems for helpers you can share and schedule.
  • ChatGPT (OpenAI). Projects that keep a job’s files and instructions together. Custom GPTs with up to 40 reference files.
  • Grok (xAI). Skills that write Word, PowerPoint, Excel and PDF files. Projects. Automations on a timer.
  • Claude (Anthropic). Long, careful writing and code, with Projects that hold context and Artifacts for documents you can keep editing.

Pick the thinking level

  • Auto usually means the fast tier. Flash, GPT 5.4 or Grok Fast: fine for rewording, summaries and quick lookups.
  • Switch up when it has to be right. Pro, GPT 5.6 Terra or Grok Expert for comparisons, policy questions, planning, and anything you will sign.
  • If the answer looks thin, open the picker, choose the careful tier, ask the same thing again.
  • Slower is not verified. A careful model still needs the source and your check.

Match the job

  • A long document or a research question: Gemini, Deep Research on, “cite the source and section”.
  • A memo or package from your files: any tool; a project keeps the files together.
  • A helper the team reruns: a Gemini Gem, a custom GPT, or a Grok or Claude project.
  • A real Word or PowerPoint file back: Grok with a skill. Budget minutes; treat the file as a start.
  • The same prompt every Monday: Grok Automations, or a scheduled Gemini Gem.

Take away

Gemini for research and shareable helpers; Grok for files and schedules; ChatGPT and Claude for document and writing work. Start fast, switch up when it matters.