Skip to content

Mastery

Last verified: 2026-05-06 · Drift risk: medium

Master the AI app you already have. Each track teaches every major capability of one provider, from the first chat to the last byte of API code, with a guided exercise at the end.

If you don't know which AI to pick, start with Any AI tool — the universal method works in almost any chat product.

Pick your track

  • Claude

    Step-by-step setup for beginners, then Projects, Memory, Artifacts, Skills, connectors, Claude Code, and API habits.

    → Open Claude

  • ChatGPT

    Chat → Custom Instructions → Projects → Custom GPTs → Tasks → Codex (cloud) → Agents SDK.

    → ChatGPT mastery

  • Gemini

    Chat → Saved info → Workspace context → Gems → Deep Research → Antigravity → Gemini API.

    → Gemini mastery

  • Grok

    Consumer Grok → Grok on X → personas → xAI API (function calling, structured outputs).

    → Grok mastery

  • Perplexity

    Ask → Spaces → Pages/Collections → Comet/agent modes → Perplexity API.

    → Perplexity mastery

  • Coding agents

    Copilot Chat → Repository instructions → Copilot cloud agent → Claude Code → Codex CLI → Antigravity.

    → Coding agents mastery

  • Any AI tool / I don't know

    A universal method that works in almost any AI chat product, plus how to switch tools when you outgrow yours.

    → Any AI tool

How each track is structured

Every mastery page uses the same shape so you can compare across products:

  1. Beginner — what to click, what to paste, how to tell whether it worked.
  2. Intermediate — projects/workspaces, saved instructions, memory, files, native tasks.
  3. Advanced — custom assistants, skills, shared workflows, evals, red-team checks, native automations.
  4. Expert — coding agents, CLI, APIs, SDKs, MCP, observability, source drift, maintenance.
  5. Level up this workflow — the exact next rung on the ladder.
  6. Guided exercise — a small task that teaches the capability by doing.
  7. Free / subscription / API availability — which features need which plan.
  8. Fallback — what to do when a feature is missing from your plan.

Capability matrix (rough comparison)

This is an opinionated, simplified comparison. Specifics drift; check the per-product mastery page for the current source-cited list.

Capability ChatGPT Claude Gemini Grok Perplexity Copilot
Plain chat ✓ ✓ ✓ ✓ ✓ ✓ (Copilot Chat)
Memory / saved instructions ✓ ✓ ✓ partial ✓ repo-level (copilot-instructions.md)
Project / workspace ✓ (Projects) ✓ (Projects) partial (Workspace context) partial (Spaces) ✓ (Spaces) the repo
Custom assistant ✓ (Custom GPTs) ✓ (Project + knowledge) ✓ (Gems) partial partial (Space + system prompt) partial (Extensions)
Skill / packaged behavior GPT Action ✓ Claude Skills (SKILL.md) Gem + files — Pages —
Native scheduled tasks ✓ (Tasks, where available) partial partial (Advanced) — — —
Coworker / agent mode ✓ (where available) ✓ Computer use (Desktop) ✓ (where available) partial ✓ (Comet, where available) ✓ Copilot cloud agent
Coding agent Codex (cloud) Claude Code (CLI) Antigravity / Gemini CLI — — Copilot cloud agent
Browser/computer use ✓ Atlas / agent (where available) ✓ Computer use partial — ✓ Comet (where available) —
Developer / API OpenAI API Claude API Gemini API xAI API Perplexity API GitHub Models

What the tracks teach in plain English

  • Chat is the floor. Every product is good at it. Most people get 80% of the value from a good prompt.
  • Memory + a saved prompt is the single biggest leverage step for non-developers.
  • Projects / Workspaces / Spaces / Gems all do the same job: "this set of instructions and files belongs together."
  • Custom assistants turn a project into a one-click thing you (and others) can launch.
  • Skills / Actions let you teach the assistant a specific procedure with templates and resources.
  • Tasks / scheduled actions automate the recurring parts. Manual-first.
  • Agent / coworker / computer-use modes drive a real surface (browser / desktop). Sandbox them.
  • Coding agents are the highest-leverage shape for code work and the fastest-changing.
  • APIs are for when your idea outgrows the app. Most readers should not start here.

When to switch tools

You start with Switch when… Switch to…
ChatGPT You want long live-document collaboration Claude Projects + Artifacts
Claude You want connectors to many SaaS apps ChatGPT Projects
Gemini You need long-form reasoning + Artifacts Claude
Grok You need structured outputs / programmatic agents OpenAI / Gemini / Anthropic API
Perplexity You want to draft long documents in the AI itself ChatGPT or Claude (writing surfaces are stronger)
Copilot Task is non-code ChatGPT / Claude / Gemini

The Surface router covers the underlying decision matrix in long form.

Plan-availability shorthand

Throughout the tracks you'll see annotations like:

  • Free: included in the free tier.
  • Sub: included in the consumer paid plan.
  • Team / Ent: team or enterprise plan.
  • API: requires developer/API access.
  • Limited rollout: announced and partially shipped; check the vendor.

If a feature is unavailable on your plan, every track ends with a fallback that works on a free or basic plan.