Start your AI journey with essential tools, basic prompt engineering, and foundational concepts. Includes reasoning / Thinking models and how supervised AI workflows change daily work.
Six short cards. One step each. Use the AI tool your employer approved.
Copilot, ChatGPT, Claude, and Gemini are strong at first drafts, summaries, restructuring messy notes, and explaining things. They are weak at facts they cannot see, numbers without a source, and final judgment.
Do this today: Mark three tasks on this week's calendar as draft, summarize, or analyze. Those are your AI candidates.
Use your work-approved tool, signed in with your work account. Never paste passwords, customer personal data, or confidential numbers into a personal account. A human checks every fact and number before anything is shared.
Do this today: Find your company's AI policy or approved-tools list and write your one-line rule on a sticky note.
An answer can come from three places: your work data, the public web, or what you attach. In Microsoft 365 Copilot, the Work IQ button controls whether it can read your work data. With it off, Copilot uses the web and what you attach.
Do this today: Before your next question, decide: my work, the web, or this file? Set Work IQ to match.
Say what you want, give the material, name the shape of the output, and ask what to verify. Four lines beat one long paragraph.
Do this today: Rewrite one prompt you used last week using those four parts. Compare the two answers.
1. Meeting notes into decisions and owners. 2. Rough notes into a status update. 3. A long thread into a short reply draft. All three are frequent, low risk, and easy to check.
Do this today: Run one of the three on real work. Time it, including your review.
Choose one that is weekly, yours, easy to check, and visible to your manager. One workflow done well beats five tried once.
Do this today: Finish this line: "I will use [tool] to [task] every [week], and track minutes and fixes."
Your beginner's starter kit: Get familiar with these tools, download the mobile apps, and create free accounts. Start with 2-3 tools that match your work needs.
These tools now support agent-style workflows: multi-step tasks, web search, and file analysis. ChatGPT, Gemini, and Claude can research, summarize, and iterate for you.
One of the most important things to understand in 2026: not all AI models think the same way. Knowing which to use saves you time and gets better results.
ChatGPT (standard), Claude Sonnet, Gemini
Optimized for fast, conversational back-and-forth. Great for drafting, brainstorming, quick summaries, and iterating on copy.
ChatGPT Thinking mode, Claude extended thinking
Think step-by-step before answering. Take longer but handle complex analysis, logic, and math much more accurately.
| Task | Chat Model | Reasoning Model |
|---|---|---|
| Draft an email | ✓ Best | Overkill |
| Complex budget model / forecast | May make errors | ✓ Best |
| Brainstorm campaign ideas | ✓ Best | Slow for this |
| Debug code / logic | OK | ✓ Much better |
| Rewrite / shorten copy | ✓ Best | Not needed |
| Multi-step analysis with stakes | Risk of errors | ✓ Best |
Read the full reasoning model guide →
Use fast chat for drafts. Switch to Think / extended thinking for analysis you would send. Use an agent when the job needs files, tools, or multiple steps.
Copy the prompt below, paste it into ChatGPT, Gemini, or Claude, and replace [TOPIC] with anything you want to learn. Great way to see how AI explains things clearly.
Current vendor-maintained learning hubs from OpenAI, Microsoft, Google, and Anthropic. No recycled old videos, no influencer filler, and no stale model-name cards.
OpenAI Academy
Official lessons on core skills, files, search, deep research, projects, custom GPTs, skills, and workspace agents.
OpenAI Academy
Role-based workflows for writing, research, data analysis, marketing, operations, finance, customer success, and managers.
Microsoft Learn
Official training on Copilot, agents, Microsoft Graph, deployment, governance, and real-world productivity scenarios.
Microsoft Adoption
Microsoft's role-based skilling center with learning paths, events, videos, and adoption guidance for Copilot users.
Google AI for Developers
Official Google guide for prototyping managed agents visually in AI Studio using templates and sandboxed agent runs.
Google Skills
Google-managed path for Gemini concepts, application building, and practical generative AI workflows on Google Cloud.
Anthropic Docs
Anthropic's current reference for prompting, output control, tool use, thinking, and agentic systems with Claude.
Anthropic Docs
Current guidance on Claude thinking behavior, effort controls, tool use, and when deeper reasoning is worth the latency.
Pick one official source, complete one lesson or workshop, then create a proof artifact from a real task. The artifact matters more than the certificate: source material, prompt or workflow, AI output, human review, and final result.
Vendor-maintained learning paths from Google, Microsoft, OpenAI, and Anthropic. These links age better than static video playlists because the vendors keep the modules current.
Once you've mastered the fundamentals, move on to AI 201 for advanced techniques, automation, and building AI-powered applications.