Lesson brief
What this module really teaches.
Models, assistants, copilots, automation, agents
The AI landscape is no longer one chat window. A modern knowledge worker now moves between general assistants, app copilots, source-grounded research, multimodal creative tools, automations, and agents.
The point of this module is not to memorize every product. The point is to build a decision habit: identify the work type, choose the lightest useful AI mode, and decide how much human review the output needs.
Most people meet AI through one chat box, so they assume the whole field is one thing. In practice, knowledge workers now use a stack: foundation models, chat assistants, workplace copilots, research systems, creative tools, automations, and agents. The first skill is knowing which layer you are using.
A non-technical learner does not need model math. They do need judgment. Is the task a draft, a search, a summary, a spreadsheet operation, a visual brief, a translation, or a workflow that touches other tools? Each answer points to a different AI pattern and a different level of review.
Futurelab field note
In Futurelab workshops, the fastest breakthrough comes when learners stop asking 'which AI is best?' and start asking 'what kind of work is this?' The same person may need ChatGPT for thinking, Gemini or Copilot inside office apps, NotebookLM for source-grounded reading, and an automation tool for repeatable handoffs.
Futurelab method
The way to do the work.
Use this as the operating pattern for the module. It keeps AI practical, teachable, and reviewable.
Start from the job
Write the actual job in plain language before naming a tool. A task like 'turn this call into a client-ready follow-up' is different from 'summarize this transcript'.
Classify the material
Text, table, slide, PDF, meeting, image, video, and app data each behave differently. Good AI use begins with knowing what kind of material you are handling.
Choose the operating mode
Use chat for thinking, search for fresh facts, deep research for synthesis, copilots for in-app work, media tools for production, automations for repeatable rules, and agents for bounded workflows.
Set the review level
Low-risk drafts can move quickly. Anything involving money, customers, law, employment, reputation, private data, or public claims needs human review before use.
Core lessons
The ideas learners must own.
These are the concepts that let non-technical learners explain what they are doing and teach it back to someone else.
The five-layer map
Model, assistant, copilot, automation, agent. A model is the engine. An assistant is a conversational interface. A copilot works inside another app. An automation follows a predefined rule. An agent can plan steps and use tools inside boundaries.
Inputs and outputs matter
Text, PDFs, spreadsheets, slides, audio, video, images, and app data each need different handling. A meeting transcript needs action extraction; a spreadsheet needs data cleaning; a visual needs a brief and rights review.
Use the lightest useful tool
Start with the simplest option that can do the job. Do not use an agent where a checklist, prompt, or spreadsheet formula is enough. Complexity should earn its place.
Operating workflow
A repeatable sequence.
Follow this order during practice. The sequence is deliberately simple so learners can remember it under real work pressure.
- 01Write the real task in one sentence.
- 02Name the input: text, table, file, audio, video, image, or app data.
- 03Name the desired output: draft, decision, chart, summary, slide, asset, or action.
- 04Choose the AI mode: chat, search, deep research, copilot, media tool, automation, or agent.
- 05Mark review level: low risk, review before use, or human approval required.
Weekly business scan
Use research mode to collect market signals, then ask for a one-page memo with facts, interpretation, caveats, and actions.
Personal admin backlog
Use a general assistant to sort errands, emails, documents, and tasks into a small next-action plan instead of asking it to do everything at once.
Team knowledge base
Use a source-grounded tool for policies and SOPs, then keep a human owner and review date on every document.
Practice lab
Build your AI landscape map
List 12 recurring tasks from work and personal life. Classify each by best AI mode, review risk, and the tool category you would test first.
Artifact fields
AI work map
- Task
- Input type
- Output needed
- AI mode
- Review level
- First tool to test
Starter prompt
I want to map where AI fits in my work. Here are my recurring tasks: [paste list]. Classify each as chat assistant, search, deep research, document copilot, spreadsheet copilot, media tool, automation, or agent. Explain the reason in plain English and flag anything that needs human review.Quality bar
What good looks like.
Before leaving the module, compare the learner artifact against these standards and common failure modes.
Tool fit
The AI mode matches the work type and is not overcomplicated.
Source clarity
The learner can explain what information the answer used and what still needs verification.
Review discipline
The output has a clear risk level and a human owner where needed.
Transferability
The learner can reuse the same classification habit on a new task next week.
Treating all AI as chat
Some tasks need sources, files, spreadsheets, design tools, or workflow permissions.
Chasing the newest tool first
A tool only matters if it changes the quality, speed, or reliability of the work.
Skipping risk classification
The same AI output can be harmless in a draft and dangerous in a client commitment.
Using agents too early
A normal prompt or checklist is often better than a complex agent workflow.
Tool categories
Tools to understand, not worship.
Current AI products increasingly blend chat, research, file understanding, app integrations, and agentic tool use. This module keeps learners grounded in categories so the curriculum does not become tool hype.
Completion
The work that proves the lesson landed.
Finish the artifact
FAQ
Questions learners usually ask.
Do I need to understand model architecture?
No. You need enough vocabulary to choose tools, understand limits, and review important outputs.
Should I standardize on one tool?
Use one default assistant, but learn categories. Documents, slides, spreadsheets, research, media, and agents solve different problems.
What is the first beginner mistake?
Asking one AI tool to do everything at once. Better work comes from choosing the right mode and reviewing in stages.