Appearance
AI Tools — A Practical Toolkit
🧰
In 30 seconds
Knowing one “great” AI tool but not when to use which still slows you down. This module isn't theory — each chapter is one tool, one real workflow: where to install it, how it's priced, what to type, and the traps that bite. Why it matters: picking the right tool for the job saves hours every day and real subscription money.
There is no “best tool” — only “the right tool for your job.” This module helps you build a personal toolkit: conversational assistants, coding agents, a dev platform, and self-hosted agents.
Tool map — which one, when
| Tool | Best at | Choose it when |
|---|---|---|
| ChatGPT | All-round assistant: write, translate, read files, analyze | Daily work, nothing to install |
| Gemini | Multimodal assistant, tied into Google's ecosystem | Inside Gmail/Docs, long context, generous free tier |
| Grok | xAI chatbot, real-time data from X | Catching trends, blunt quick answers, already on X Premium |
| Perplexity | Answer engine with citations | Research that needs sources |
| NotebookLM | Assistant that answers only from your documents | Studying/researching from your own PDFs |
| Claude Code | Autonomous coding agent in the terminal | Real projects, editing many files |
| OpenAI Codex | Coding agent — CLI / Cloud / IDE | Delegating background, parallel coding tasks |
| Cursor | AI code editor (VS Code fork), Tab & Agent | Coding by hand while delegating to AI; best autocomplete |
| Windsurf | AI-native IDE, autonomous Cascade agent | Like Cursor but want a stronger agent |
| v0 | Prompt → working React/Next web UI | Spin up UI/landing pages fast, light on code |
| Replit | Cloud IDE + agent that builds & deploys | Beginners, build in the browser, deploy instantly |
| Claude Cowork | Office/knowledge-work agent (no code) | Automating paperwork, research |
| n8n | No-code automation, drag-and-drop AI agents | Connecting apps, automated workflows, self-host |
| Notion AI | AI right inside your notes/docs workspace | Already on Notion, want AI to write & answer from docs |
| Gamma | Beautiful slides/sites/docs from one prompt | Fast presentations, not a designer |
| ElevenLabs | Most realistic AI voice, cloning & dubbing | Voiceover, podcasts, multilingual video dubbing |
| GitHub | Store code, collaborate, version control | Every coding project — a foundational skill |
| OpenClaw | Open-source agent running locally, controlled via chat | Self-host, private, wired to Telegram/Slack |
| Hermes Agent | Self-hosted agent with memory, multi-channel | A private 24/7 assistant that remembers long context |
🧭 Suggested path
- New to AI → start with ChatGPT (all-round assistant) + GitHub (foundational skill).
- Search & research → Perplexity (with sources), NotebookLM (from your own docs).
- Code with AI → Cursor (IDE) or Claude Code / OpenAI Codex (terminal/agent).
- Non-coders / office work → Claude Cowork; automate with n8n.
- Compare the big chatbots → ChatGPT vs Gemini vs Grok.
- Your own self-hosted agent → OpenClaw, Hermes Agent.
🔗 Combining tools — a real workflow
Real power isn't in one tool — it's in chaining them. For example, taking a web-app idea from zero to shipped:
| Step | What you do | Tool |
|---|---|---|
| 1. Idea & spec | Brainstorm, write a PRD, sketch screen flows | ChatGPT |
| 2. Build code | Hand over the spec, let the agent scaffold & edit files | Claude Code or Codex |
| 3. Store & collaborate | Commit, branch, Pull Request, review | GitHub |
| 4. Background automation | An agent runs on a schedule, scans issues, reports back | Cowork / Hermes / n8n |
Tip: use one conversational assistant to THINK (ChatGPT/Claude) and one coding agent to BUILD (Claude Code/Codex) — don't make a single tool carry everything.
Free or paid?
| Tool | Free tier? | When you'll need to pay |
|---|---|---|
| ChatGPT | (limited) | Stronger models, heavy use, Projects/Agent |
| Gemini | (generous) | Pro for top models, Deep Research, long context |
| Grok | (limited) | SuperGrok / X Premium for heavy use |
| Perplexity | (limited) | Pro ~$20: Deep Research, stronger models, Comet |
| NotebookLM | (generous) | Plus when you need more notebooks/sources |
| Cursor | (limited) | Pro ~$20 with heavy use (credits run out fast) |
| Windsurf | (limited) | Pro ~$20 with heavy use |
| v0 | (monthly limit) | Pay-as-you-go (tokens) when building a lot |
| Replit | (limited) | Core ~$20 + compute when deploying |
| Claude Code | (paid plan/API) | From the start (Pro/Max or API) |
| Claude Cowork | (within Claude plan) | With a paid Claude plan |
| OpenAI Codex | (with a ChatGPT plan) | Heavy use / large cloud runs |
| Notion AI | (add-on) | ~$8–10/user/month (bundled into Notion) |
| Gamma | (400 credits) | Plus/Pro for more decks / no watermark |
| ElevenLabs | (character limit) | Paid for more voice minutes / cloning |
| GitHub | (broad) | Team features / Copilot Pro (students: free) |
| OpenClaw | (MIT — free software) | Only LLM tokens (or $0 fully local) |
| Hermes Agent | (MIT — free software) | Only tokens (or $0 with Ollama/free tier) |
Quick FAQ
Total beginner — which tool first?
ChatGPT (all-round assistant, nothing to install) + GitHub (foundational skill). These two cover almost everything and are mostly free.
Don't code, want AI to handle paperwork/research?
ChatGPT for one-off tasks; Claude Cowork when you want to hand over a multi-step goal and get a finished deliverable.
Want AI to write code for a real project — Claude Code or Codex?
Pick by the plan you already have: paying for Claude Pro/Max → Claude Code; paying for ChatGPT → Codex. Both are strong — the difference matters less than how disciplined you are with them.
Worried about data/privacy, don't want to send things to the cloud?
Use a self-hosted agent (OpenClaw / Hermes Agent) with a local model (Ollama) → your data never leaves your machine.
A note on timeliness
AI tools change pricing & features fast. This content is current as of around mid-2026 — always check the official site before buying a plan.