In Claude, go to Settings → Profile. Tell it who you are, what you do, how you communicate, what context it should always have. Think of it as the brief you'd give a new hire on day one. Most people skip this and it makes a huge difference — every conversation starts smarter.

On Mac: turn on dictation (System Settings → Keyboard → Dictation). Press the microphone key or double-tap the fn key, and start talking. It transcribes in real time. This is a game changer for getting thoughts into Claude quickly.

Instead of typing a long prompt, just talk. "Hey Claude, I need to put together a project plan for the Q2 vendor review, here's what I'm thinking..." Your spoken thoughts are often better context than what you'd bother typing out. Then let Claude clean it up.

After Claude produces something important, ask it to review its own work. This sounds weird but it catches a lot. Especially important for anything with numbers, dates, or factual claims.

Get in the habit of asking these follow-ups after any important output:

Most people get a mediocre first output and either give up or start over. Don't. Stay in the same conversation and keep refining. Even if you could make the edit faster yourself in a Google Doc, keep giving Claude feedback so it gets better and learns you better. The time you sink now into doing this will pay off in every future conversation.

Think of it like training a new teammate. The first week is slow. By week three, they're anticipating what you need.

The single biggest lever you have. AI isn't bad at its job — it's working with whatever you give it. You don't need a perfect brief. Just give it strong context: who you are, who it's for, what you need, what tone, and any examples or constraints.

When Claude delivers something, don't give feedback one thing at a time. Go through ALL of it and send everything at once. Bullet point what you want changed. This is faster and produces way better results because Claude can see the full picture.

Don't worry about typos. Claude knows what you mean. And when Claude asks you to do something yourself ("you could go to Settings and change X"), say "can you just do it?" Don't accept unnecessary homework.

Don't tell Claude "first do X, then do Y, then do Z." Just tell it the problem. The way you'd do something isn't necessarily the way AI would do it — its capabilities are different. Explain big problem statements: your pain, your expectations, what the end result should look like. Let it figure out the how.

Before any meeting, project, or conversation, ask Claude to get you up to speed. Even on stuff you already know. Meeting with the Content Marketer? Quick "remind me what's important to them from me, the buckets they think about their job in, and based on their priority stack what should I hit on."

Instead of describing what you want from scratch, give Claude a URL. It can read websites and pull context directly.

One of the most underrated things AI can do is make things look good. You already have the content — the SOPs, the playbooks, the docs buried in your Drive. Nobody reads them because they're a wall of text in a Google Doc from 2023.

If you want people to actually use the thing you made, it needs to be easy to scan, nice to look at, and feel like someone cared. AI makes that trivially easy. This site you're reading right now? Built entirely with Claude.

Stop copy-pasting paragraphs. Upload the actual file: the PDF, the spreadsheet, the doc, the screenshot. Claude can read all of them. The more raw material you give it, the better the output. This is especially important when you're first getting started — over-share, not under-share.

Before you ask Claude to make anything, ask it to think with you. And use it even for things you've done a hundred times. I've done this with RevOps stuff I could do in my sleep and still gotten a new way to frame it.

Don't treat Claude like a search engine. Don't treat it like a fragile employee. Treat it like a smart, direct colleague. If the output is bad, say "this is bad." If it's bland, say "this is bland and doesn't speak to my company at all." You can be vague — "this doesn't feel right" is valid. You don't need to diagnose the problem to give good feedback. Claude won't get offended.

When you start connecting tools, you'll have API keys and tokens floating around. Do not copy-paste these into notes, Slack messages, or shared docs. Use a password manager.

Any password manager works (Bitwarden, Dashlane, LastPass). 1Password is worth knowing about because it has AI-native capabilities: it can auto-fill API keys into developer tools and manage secrets across your workflows. If your company already uses one, use that.

Rules: Never paste API keys into a chat window. Never put them in a shared Google Doc. Never send them over Slack. Store them in your password manager and access them from there.