Tau
AI at work: the essentials · 12 lessons

How to use AI at work, whatever your role

Learn which AI tools you can use at work, what's safe to put in them and how to check AI's work before it goes out in your name. For every job, and it helps employers meet AI training rules such as the EU AI Act.

Who it’s for: Every employee, in any role, industry or level of experience.

Plenty of people now use AI at work, often without anyone showing them how. That's where mistakes creep in. Someone pastes a client list into a free chatbot, or sends a report with a figure the AI made up. This course teaches the habits that prevent those mistakes, whatever your role.

You'll start by finding your organisation's approved AI tools and policy. Then you'll sort information with a simple traffic-light system. Green information is public. Amber is internal and goes only into approved tools. Red stays out unless a system is approved for it. From there, you'll draft emails you'd happily sign, summarise long documents with quotes you can check, and turn meeting notes into a list of decisions and who is responsible for each.

You'll also learn a two-minute check to run before work leaves your desk. Other lessons cover spotting AI-written phishing and hidden instructions in files, and knowing when AI is the wrong tool for the job. The course helps employers meet AI training rules such as those in the EU AI Act. In your final project, you'll measure how much time AI really saves on a task you do every week.

By the end you’ll be able to…

  • Sort any work information into what may go into which AI tool
  • Draft, summarise and analyse with AI, then verify before it leaves you
  • Spot AI-made phishing, deepfake requests and hidden prompt instructions
  • Measure the net time AI saves on one real task in your job

Short on time? Start with the essentials

These 4 lessons cover the most useful ideas in this course. Each one takes about a minute.

Essential 1

Use your organisation's approved AI tools and policy

Find your organisation's approved AI tools and AI policy, and explain why using unapproved 'shadow AI' puts company data and you at risk.

Essential 2

Sort work data with a red, amber, green traffic light

Classify information before it goes into any AI tool: green is public, amber is internal and only for approved tools, red stays out unless a system is approved for it.

Essential 3

Draft an email with AI and own every word

Draft an email or document with AI from your own notes, then edit facts, numbers, promises and tone until you would sign it without the AI.

Essential 4

Run a two-minute check before work leaves your desk

Apply a short checklist to AI-assisted work (facts, numbers, names, sources, confidential details, tone) before it reaches a client, manager or the public.

Every lesson in this course

This course has 12 video lessons of about a minute each, from the basics to expert habits. Each one ends with a question, three takeaways and its sources.

Foundations 2 lessons

  1. Use your organisation's approved AI tools and policyEssential

    Find your organisation's approved AI tools and AI policy, and explain why using unapproved 'shadow AI' puts company data and you at risk.

    “Same chatbot, two logins: one keeps work data under contract, one may not.”

  2. Sort work data with a red, amber, green traffic lightEssential

    Classify information before it goes into any AI tool: green is public, amber is internal and only for approved tools, red stays out unless a system is approved for it.

    “Green: the press release. Amber: this quarter's sales. Red: a colleague's sick note.”

Building skills 7 lessons

  1. Draft an email with AI and own every wordEssential

    Draft an email or document with AI from your own notes, then edit facts, numbers, promises and tone until you would sign it without the AI.

    “Your name goes on the email, not the AI's. Three edits before you hit send.”

  2. Turn a meeting transcript into decisions and owners

    Turn meeting notes or a transcript (recorded with everyone's consent) into decisions, actions, owners and dates, and check each against what was actually agreed.

    “60 minutes of transcript → 5 decisions and 7 actions, each with a name and date.”

  3. Ask a spreadsheet a question and check the numbers

    Ask AI a plain-English question about a spreadsheet, then confirm which rows, filters and formula it used and recalculate one figure yourself.

    “'Sales up 12%.' Which rows did the AI count? Check before you quote it.”

  4. Summarise a long document with quotes you can check

    Summarise a long report, policy or contract so each key point carries a direct quote and page or section reference you can find in the source.

    “A 60-page report in 8 bullets, each with a quote you can find with Ctrl+F.”

  5. Run a two-minute check before work leaves your deskEssential

    Apply a short checklist to AI-assisted work (facts, numbers, names, sources, confidential details, tone) before it reaches a client, manager or the public.

    “Lawyers were fined for citing cases AI invented. A two-minute check would have caught it.”

  6. Tell colleagues and clients when AI helped

    Decide when and how to disclose AI use, following your organisation's policy and client expectations, in one clear line.

    “'Drafted with AI, checked by me.' When that one line keeps trust, and when it's required.”

  7. Spot AI phishing and hidden instructions at work

    Recognise AI-written phishing, deepfake voice or video requests, and prompt injection hidden in emails or files your assistant reads, and know how to verify and report them.

    “A video call with the 'CFO' and colleagues. All deepfakes. US$25 million sent.”

Advanced 2 lessons

  1. Know when AI is the wrong tool for the job

    Recognise tasks where AI should not lead (decisions about people, regulated advice, sensitive negotiations, tasks it is weak at) and choose what to do instead.

    “Consultants using AI were 19 points less accurate on the one task AI was bad at.”

  2. Build a five-prompt library for your own role

    Save five tested prompts for your recurring tasks, each with context slots, an output format and a built-in check, where your team can reuse them.

    “You retype the same prompt every week. Save it once, with the check built in.”

Expert 1 lessons

  1. Write a one-page AI working agreement for your team

    Agree with your team which tasks use AI, with which tools, who checks what, and how AI use is disclosed, in one page that sits under company policy.

    “Half the team uses AI, half doesn't, and nobody knows who checked what.”

Apps we recommend

These are the tools this course uses, and what each one is best for. We have no partnership with any of them.

Microsoft 365 Copilot Chat

Best for: Organisations on Microsoft 365

Secure AI chat for Microsoft 365 work accounts, under your organisation's existing data protection terms; no extra cost on eligible plans.

Freemium

Gemini in Google Workspace

Best for: Organisations on Google Workspace

Gemini inside Gmail, Docs, Sheets and Meet; Google says company data is not used to train its models or for ads.

Paid

ChatGPT Business

Best for: Teams that need a managed ChatGPT workspace

Shared workspace with admin controls and SSO; business data isn't used to train OpenAI models by default.

Paid
Hands-on project

Two weeks of AI on one recurring task

Pick a task you do every week, such as a status report, inbox triage or meeting notes. Time it without AI, then do it with your approved AI tool for two weeks, timing drafting, checking and fixing separately. Use anonymised data where needed, then decide whether to keep, change or drop the new way.

What you’ll make: A one-page before/after time log with net minutes saved, your verification checklist, and a keep, change or drop decision

Check your own work

  • Baseline times are recorded for at least 3 runs of the task before AI is used
  • AI-assisted times include checking and fixing, not just drafting
  • Every run used an approved tool and no red-category data
  • The checklist names at least 4 specific checks and was applied to every output
  • The decision cites net time saved and at least one error the checks caught or ruled out

Further reading and viewing

We’ve chosen these articles, research papers, videos and guides from other trusted sources.

Guide · European Commission · 15 min

AI Literacy - Questions & Answers

The Commission's own answers on what the EU AI Act's AI-literacy duty asks of employers, including recent changes.

Research · Harvard Business School (Dell'Acqua et al.) · 30 min

Navigating the Jagged Technological Frontier

758 BCG consultants: AI raised quality on some tasks and cut accuracy on one it was bad at.

Course · Microsoft Learn · 45 min

Work smarter with AI

Free beginner path on grounding, agents and responsible AI use for everyday workplace tasks.

Article · OWASP Gen AI Security Project · 10 min

LLM01:2025 Prompt Injection

The security community's reference on how hidden instructions in content hijack AI assistants.

Questions about this course

Is it safe to use ChatGPT at work?

It depends on the tool, the version and your employer's rules. Many organisations approve specific tools, such as a business version with data protections, and ban others. Before you paste anything in, check your AI policy and sort the information. Public material is usually fine. Client, personal or confidential data should only go into tools approved for it.

What should I never put into an AI tool at work?

Unless a tool is approved for it, keep out customer and employee personal data, passwords and access keys, unreleased financial figures, legal or commercial secrets, and anything covered by a confidentiality agreement. A simple traffic-light system helps. Green is public, amber is internal, and red stays out. When you're unsure, ask your manager or IT team first.

How do I check AI's work before I send it?

Run a short check every time. Confirm facts, numbers and names against the source. Open every link and reference to make sure it exists and says what the AI claims. Remove confidential details, and read the tone as the recipient would. If you wouldn't sign it without the AI, it isn't ready. One lesson turns this into a two-minute habit.

Does the EU AI Act require AI training for staff?

The EU AI Act asks organisations that provide or use AI systems to take steps so their staff understand AI well enough for their role. It applies to organisations within its scope, including some based outside the EU. This course gives every employee the essentials, and it can support that duty. Check with your own compliance team what applies to you.

How long are the lessons?

Each lesson takes about a minute: a short video, one question, then three takeaways and the sources. You can fit one into a coffee break. A three-question skills check sets your level so you start in the right place. The first 5 lessons are free, and Tau Pro gives you every course, including the profession courses.

Get early access to this course

Tau opens soon. Join the waitlist and we’ll email you as soon as you can start this course.