Create your account
What an account gets you that trying a tool without signing in doesn't, work accounts versus personal ones, and how to sign up.
Set your privacy settings
What it means when a tool learns from your conversations, where the setting lives if the tool has one, and what turning it off does and doesn't do.
Your own customization and safeguards
The setting that carries your instructions into your conversations, and how to use it to reduce some common ways AI output goes wrong.
Variety of AI tools
The main kinds of AI tool, what each is built around, and questions that help you work out which fits what you're doing.
Free versus paid
What paying for an AI tool actually changes, and how to tell whether the free version is limiting you.
Temporary and incognito chats
What a temporary chat window is, when to reach for one, and what it does and doesn't protect.
Advanced customization
Getting more out of the standing-instructions field — what to put in it beyond tone, and how to tell whether it's working.
Saved memories
What the AI quietly remembers about you, how to see it, and why it may be the reason answers feel off.
Shaping your AI assistant over time
Steering the AI inside a conversation — telling it what worked, correcting it, pushing back when it seems wrong — and what carries forward versus what resets when you start a new chat.
Scheduled prompts
Having the AI run something on a schedule rather than when you ask — a morning briefing, a weekly check — what it is good for, and what to watch when it runs without you.
Projects and notebooks
The feature that keeps instructions, files and past chats together for one piece of ongoing work — called projects in Claude, notebooks in Copilot, and other names elsewhere — including sharing one with collaborators.
AI built into the software you already use
Copilot inside Word, Gemini inside Google Workspace, assistants embedded in the apps you work in daily — how they differ from a chat window, what they can already see, and when to reach for one instead.
AI workspaces
Handing the AI a whole task rather than asking it a question — tools that work in a file space over time and hand back finished work.
When it isn't working
The common ways a session goes wrong — refusals, repetition, forgetting what you attached, contradicting itself — what is actually causing each, and what to do about it.
Usage limits and how to stretch them
Why you hit a cap mid-task, what actually burns through your allowance faster, and the habits that get more out of the same plan.
Talking to AI out loud
Voice mode as a genuinely different way to use AI — thinking aloud, rehearsing a conversation, hands-free questions — and what it isn't good for.
Using AI as a thought partner
Working through your own thinking with the AI — brainstorming, pressure-testing an idea, finding the hole in your reasoning — where the value is the thinking, not the output.
Using AI for research
Finding, gathering and synthesizing information across sources — what the AI is genuinely good at here, and where the answer needs checking before you rely on it.
Using AI as a writer or editor
Drafting from a blank page and improving what you have already written — two different jobs, with different instructions and different risks to your own voice.
Using AI to manage your email
Drafting replies, triaging an inbox and summarizing threads — how to actually do it, and the risks that come with giving an assistant access to your mail or letting it send on your behalf.
Using AI to manage your calendar
Scheduling, rescheduling and making sense of a busy week — how to set it up, and what to check before letting an assistant move things or accept invitations for you.
Using AI to fix things
Photographing something broken — an appliance, a leak, an error message, a part you cannot name — and working out what it is and what to do about it.
Using AI for meetings
Briefing yourself beforehand, and what AI notetakers do and don't capture — plus the consent question most people skip before switching one on.
Using AI for slides and presentations
Getting from an idea to a deck — structure, narrative and speaker notes — and avoiding the bullet-point soup that AI decks default to.
Using AI with spreadsheets and data
Writing and explaining formulas, restructuring messy data, and finding the pattern in a sheet — plus why to check every number it gives you.
Decoding a document
A contract, a bill, an official letter, a policy — working out what it actually says, what it means for you, and what to ask about.
Learning something new with AI
Using AI as a tutor rather than an answer machine — explanation before answers, being quizzed, finding your own gaps — and the trap of mistaking a good explanation for learning.
Saving and sharing prompts
Keeping the prompts that work so you can use them again, and handing them to other people — including why the same prompt may not behave the same way for someone else.
From a single prompt to an AI system
What people mean by an AI system, and the rungs between typing one prompt and having something that runs as a system — written for someone who has never built anything.
Building a grounded AI system
What it actually looks like to build something that answers from a particular body of material — the knowledge base you assemble, the context layer that decides what reaches the model, and the prompt that sets the task.
Prompt basics
Specificity, context, stating what “good” looks like, expecting to iterate.
Let the AI interview you
Having it ask you questions first — one at a time — before producing anything.
Ask the AI to write your prompt
Having it draft the prompt itself, then testing and revising until reliable.
Keeping your output from reading as AI slop
Why AI writing comes out generic, padded and hedged, and how to instruct your way out of it — keep it short, cut the throat-clearing, drop the standard three-part structure.
Asking the AI to rank its own answers
Asking for several options and having the AI rank or critique them rather than taking the first answer — what this is genuinely good for, and the limits of a model judging its own work.
Human in the loop
Why review matters, why it erodes under deadline pressure, how to make it survive.
AI bias and discrimination
Where bias comes from, where it surfaces, and where it matters most.
Made-up facts and the verification habit
It invents facts, sources, quotes and links that look real — and how to catch them.
What AI can and can't do well
Strengths, weaknesses, and the judgment call about whether to use it at all.
Context windows
The limit on what it holds at once, and how it loses track without telling you.
Recognizing AI-written work
The tells in something someone has handed you — and what spotting them should and shouldn't lead you to conclude about the person who sent it.
The AI's file space
Generated files sit in the tool's storage, not yours, until you move them.
What you're granting with access
Connecting a folder grants more — and for longer — than it appears.
When AI acts on your behalf
Read vs. write; once it can send or change things, errors escape the chat.
The threat of prompt injection
A page or document the AI reads can carry text meant for it rather than for you — what that looks like, why tools fall for it, and when it starts to matter for you.
When AI behavior changes
Models update silently; prompts that worked last month stop working.
Grounding and retrieval
It searches your documents selectively rather than reading all of them.
Connectors
What a connector mechanically is; read vs. write; persists until revoked.
Agentic and non-agentic AI
How many steps run without you — and cutting through vendor “agent” marketing.
What the AI doesn't know about itself
Ask a model which version it is, where its knowledge ends, or why it said something — and the answer can be confidently wrong. It has no privileged view of its own workings.
How AI produces an answer
It predicts what comes next, one piece at a time, based on what is likely to follow. Almost everything surprising about these tools comes from that one fact.
Rolling AI out to your team: the policy basics
Settling tool, accounts, data boundaries and disclosure for a group.
Ground rules and etiquette with the people you work with
Settling what gets disclosed, whose material is in bounds, and what nobody should be doing — plus the unwritten etiquette of using AI around colleagues.
AI and kids
Family accounts, homework, and where AI genuinely helps a child learn versus where it does the thinking that was the point.
Managing confidentiality risk
What drives exposure, and why account choice is the single biggest lever.
Who owns what AI makes
Your rights in AI output, and whether output might infringe. Genuinely unsettled.