← Back to AI Basics
AI BASICS · CHOOSING AI TOOLS
BeginnerLevel 1

How to choose the right AI tool.

Start with the job, not the tool. A simple process for deciding whether ChatGPT, Claude, Codex, or another AI tool fits the work in front of you.

6 minTested · July 7, 2026
Watch the walkthrough

AI Basics - How to choose the right AI tool

5:36

Overview

By the end of this page, the goal is not to know which AI is "best" overall. The goal is to have a simple process for deciding which AI tool fits the job in front of you.

The common question is usually too broad: should I use ChatGPT, Claude, Codex, Copilot, Gemini, Perplexity, or something else? Those tools overlap, but they are built for different kinds of work, and the same app can run different models with different plans and access.

So the process starts with the job, not the tool. Name what you are trying to produce, what the AI needs to see, and whether it should only advise or actually do work somewhere. Then check your constraints and route the job. That is the whole skill.

What you need

  • A specific job or task you want AI to help with.
  • Access to at least one general AI assistant, such as ChatGPT or Claude.
  • The ability to check whether the job needs files, current sources, coding access, or just a normal chat. Some workplaces restrict these, so know what you are allowed to connect.

Instructions

  1. 01Step

    Know what you are choosing between

    Instead of asking, "Which AI is best?", ask, "Which tool fits the job I am trying to do?" ChatGPT, Claude, Codex, Gemini, Copilot, and Perplexity are all useful, but for different kinds of work.

    Before comparing them, know the four things people usually mix together:

    • The app: where you do the work, like ChatGPT, Claude, or Codex.
    • The model: the engine inside the app. ChatGPT can run different GPT models, and Claude can run Haiku, Sonnet, Opus, or Fable. The same app can feel different depending on the model.
    • The workspace: a normal chat, a project, connected files, a coding workspace, or a work app like Word or Excel.
    • The access: free plan, paid plan, work account, usage limits, and what the tool is allowed to see or touch.

    You do not need to memorize this list. Just know these are four separate choices, so when someone says one AI is better than another, you know which of the four they are actually talking about. To see the current lineups, check the OpenAI model docs, the Codex model docs, and the Claude models overview.

  2. 02Step

    Name the job

    This is where my actual decision starts. Before I open anything, I say the job in one sentence: what am I trying to produce?

    "Write a follow-up email." "Compare three vendors for my boss." "Fix a bug on my website." "Understand this contract."

    Then two quick follow-ups: what does the AI need to see (notes, files, a spreadsheet, a project folder, the live web), and where does the work end up (a document, a spreadsheet, a website, or just an answer)?

    If you cannot say the job clearly, switching from one chatbot to another will not fix the output. The tool matters, but the job comes first.

  3. 03Step

    Decide: advice or action?

    This is the biggest fork in the road. Does the AI just need to give you words, or does it need to do work somewhere?

    • Advice: the AI gives you a draft, an answer, a plan, or feedback, and you take it from there. A normal chat handles this well.
    • Action: the AI needs to open files, edit a project, run checks, or work inside another app. That takes a tool with real access, like Codex, Claude Code, or a connected work app.

    Most bad tool choices come from mixing these up: asking a plain chatbot to do file work it cannot see, or setting up a heavy coding agent for a question a chat could answer in seconds.

  4. 04Step

    Check the constraints

    Once the job is clear, the constraints are how I narrow the list to one or two tools. Here is what I check:

    • Cost and usage limits: smarter models usually cost more. If you hit the limit in 10 minutes, it is not the right tool for that job.
    • Data sensitivity: client files, company data, and personal or financial details should change what you paste or connect.
    • Connections: email, files, project folders, and workspaces make a tool more useful, but they also raise the review bar.
    • Collaboration: a team may need shared projects and repeatable instructions instead of one person's private chat history.
    • Verification: for current facts or high-stakes work, the last step is still checking sources and reviewing the result.

    Different tools are built for different constraints. Some let you mix expensive and cheap models, some give you real data controls, some show sources, and some are built for approvals before they act. That is why the same tool can be the right pick for one job and the wrong pick for the next.

  5. 05Step

    Route the job

    Now pick the tool. Here is my current routing, with the reason each route works:

    • Drafting or writing in a natural voice: Claude, because its writing needs the least editing for my taste.
    • Thinking through an idea, brainstorming, or planning: ChatGPT or Claude, because they are simple to interact with and give you access to a lot of intelligence.
    • Reviewing documents: Claude, and Codex when the review has to happen across many files, because Codex can open the files itself.
    • Code and website work: Codex and Claude, because the work lives in project files the AI needs to open, change, and check.
    • Working across apps: Codex, because its plugins let it reach more of my apps from one place.
    • Current research and fact-checking: Google AI Mode or Perplexity, because they search real sources instead of answering from memory.
    • Word, PowerPoint, and Excel work: Claude or ChatGPT, because I prefer their output and their add-in paths over Copilot for these jobs.

    Your routes can be different from mine. The reasons are the part worth copying: match what the job needs to what the tool can see and do.

  6. 06Step

    Walk one real job through it

    Here is the whole process on one real job. Your manager asks you to look into three project-management tools and bring back a recommendation.

    • The job: a one-page comparison your manager can decide from.
    • The inputs and destination: current pricing and features from the vendors' sites, ending up in a short document.
    • Advice or action: advice. You need research and a draft, not file changes.
    • The constraints: pricing changes often, so the research must come from current sources, not chatbot memory. And it is work output, so check the facts before sending.
    • The route: Google AI Mode or Perplexity to gather current facts with sources, then Claude or ChatGPT to turn the notes into a clean one-pager.

    One job, two tools, no guessing. And if the pricing lives in a folder of vendor PDFs instead, the job changes: now the AI needs to open files, so I would reach for a tool with file access.

  7. 07Step

    When a new tool shows up

    You do not need to write any of this down or build a system. After you run a few jobs through these questions, the routing takes about ten seconds and happens in your head.

    The questions are the part that lasts. Models change, plans change, and new tools launch every month. What I take into consideration stays the same: what am I producing, what does the AI need to see, advice or action, and what constraints apply.

    That also tells you what to do when a new tool or model shows up. Do not start over. Pick one job you already understand, run it through the new tool, and see if it beats your current pick. If it does, it earns the spot. If not, you can ignore the hype.

    And if you want a shortcut on any single job, the prompt below asks an AI to route it for you. It gets a little meta, but it works.

The prompt

Ask an AI to help you pick the AI

I need help choosing an AI tool.

My job: [what I am trying to accomplish]
What the AI needs to see: [files, data, notes, browser, project files, app, or source material]
The output I want: [draft, answer, plan, spreadsheet, code change, research summary, etc.]
My constraints: [free plan, usage limits, privacy, work data, connected apps, team sharing, deadline]

Recommend the kind of AI tool that fits this job and explain the tradeoff in plain English.