9 Claude Tips That Put You Ahead of 99% of Users 🧠
Most people use Claude at a fraction of its potential. Here are nine quick tips that'll get you noticeably more efficient and effective results from your day to day use.
Tip 1: Set Up Custom Instructions 📝
Custom instructions tell Claude who you are, how you want it to behave, and what your preferences are before you've even typed a message. Things like your job, your goals, the tone you want, and the little details that matter to you.
If you don't know where to start, just ask Claude to build them for you:
"I want to create some custom instructions to go into settings. Can you help? Ask me questions."
It'll ask a few follow up questions and build a solid set you can drop straight into settings.
A couple of extra tips:
- ✅ Tell it your language preferences. For example, always British English
- ✅ Flag anything you hate, like em dashes in outputs. Say so once and Claude respects it consistently from then on
If you're coming from ChatGPT or another AI chatbot, you can use the memory export feature to pull out what it already knows about you. Go to Settings > Capabilities > Import memory from other AI providers.
Tip 2: Use Extended Thinking for Complex Questions 🤔
In supported conversations, you can ask Claude to think deeply before it responds. Just say "think carefully about this" or "use extended thinking" before a complex question.
Claude will reason through the problem step by step before giving you an answer. The difference in output quality on hard problems is noticeable.
Not needed for every task, but for anything involving strategy, analysis, or decisions with lots of moving parts, this is the one to use.
Tip 3: Set a Writing Style ✍️
This one's underrated. Claude lets you set a writing style, and it's surprisingly useful for any kind of content creation.
You can upload examples of your own writing or choose from preset styles, and Claude will match that tone in its output. So whether you're writing scripts, emails, or social posts, it's not just producing generic AI-sounding text. It's trying to sound like you.
Click the plus sign at the bottom of a chat, go to Use Style, and you'll see generic presets plus the option to build your own. Genuinely one of the most underused features in Claude.
Tip 4: Build Interactive Apps Inside the Chat 🛠️
This one genuinely surprised me the first time I used it. Claude can build fully functional interactive apps right inside the conversation.
I asked it to build a YouTube script scoring tool and described what I wanted. It built a live interface, and now I can paste any script draft into it and get an instant quality score based on hook strength, pacing, and CTA placement. All inside Claude. No external app, no code, no subscriptions, no browser.
You could build a habit tracker, a decision-making framework, a content calendar generator, a client proposal builder. Whatever your workflow needs, describe it and Claude builds it inside the app.
Most people use Claude like a search engine. This is where it starts feeling more like a personal software developer that works for free.
Tip 5: Edit Instead of Resending 🔁
A small habit change that makes a big difference. Instead of sending a new message when Claude gets something wrong, hover over your original message and hit edit. Change what you need and resubmit, and Claude rebuilds the response from that point.
You get a clean branch in the conversation without a trail of back-and-forth corrections. It keeps things tidy, and your results genuinely get better, faster.
Tip 6: Pick the Right Model for the Job 🎛️
Claude has different models, and picking the right one matters, but hardly anyone actually does it.
Sonnet is the default and it's great for the vast majority of tasks. Fast, capable, handles most things well.
For something complex, a big strategic decision, a detailed analysis, or a long document that needs real depth, switch to Opus. It's slower but noticeably more thorough.
Spending two seconds choosing the right model for the job is a genuinely quick win.
Tip 7: Turn On Web Search 🔍
Claude can search the web too, but it's not always obvious the feature is active. Check that web search is toggled on in your conversation.
Once it is, you can ask about current events, recent releases, live pricing, anything that would normally sit outside its training data. Claude pulls in real sources and tells you where the information came from.
Sometimes it activates automatically, sometimes it doesn't. Worth double checking, particularly for anything time-sensitive.
Tip 8: Use Deep Research for the Big Stuff 📊
At the opposite end of the scale from a quick web search is Deep Research, found as an option when you start a conversation.
Give Claude a research topic and it'll pull from multiple sources, cross reference them, and come back with a structured report. The kind of output that would normally take you hours to put together yourself.
It takes a few minutes to run, but the depth of what comes back is genuinely impressive. Use it for competitive research, topic deep dives, and anything where you need more than a quick answer.
Tip 9: Use Projects So You're Never Starting From Scratch 🗂️
Most people treat every Claude conversation like a blank slate. Projects completely changes that.
Create a project, give it custom instructions specific to that context, and upload documents Claude can reference across every conversation inside it. Every chat inside that project starts with Claude already knowing all of that context. You're not re-explaining yourself every time. You're getting straight to the useful stuff.
I've got a project set up for this channel with my style guide, video format, and examples of past scripts. When I ask Claude to write something for Feel Productive inside it, Claude already knows exactly what that should sound and feel like.
I still tweak and change things myself, but it gets me about 90% of the way there.
The Bigger Picture
None of these tips are complicated. Custom instructions, the right model, a writing style, a project set up properly. Small, deliberate choices that compound into noticeably better output every single time you open Claude.