I’ll be honest. The first time I tried Claude, I closed the tab after about ten minutes and went back to my old habits. A lot of people I know did the same. So a post from an AI professional on LinkedIn caught my eye with one blunt line: “You’re looking at Claude in 2023.” If your opinion of Claude comes from what it could do a few years ago, that picture is very out of date.
The author put together a short roadmap of eight areas to learn so you can keep up with how much the tool has changed. I love how practical it is. Every item fixes a real frustration, from weak prompts to hitting usage limits. Below I go through each point, add some context on why it matters and show how you can use it this week.
1. Set Claude Up Properly Before You Judge It
The creator says to start here “if you’ve tried Claude twice and decided it was useless.” That line describes a lot of people. Most bad first experiences come from using Claude like a plain chat box with no setup at all.
Spending a few minutes on setup makes a big difference. Here’s what I’d focus on:
- Profile preferences: Tell Claude who you are, what you do and how you like answers formatted.
- Projects: Group related chats and upload the reference files Claude should always have.
- Styles: Set a default tone so you don’t have to repeat “keep it short and casual” every time.
Once Claude knows your context, the answers stop sounding generic and start sounding like they were written for you.
2. Learn Prompt Engineering (For Real This Time)
This is my favorite line in the whole post: “make this better” is barely an instruction. It’s true. Claude can’t read your mind, so vague requests get vague results.
A good prompt usually covers four things: the goal, the audience, the format and the limits. Compare “make this email better” with this:
“Rewrite this email for a busy CFO. Keep it under 120 words, lead with the cost savings, and end with one clear yes-or-no question.”
The second version gives Claude a target to hit. Add an example of the output you want and quality jumps again.
3. Use Opus 5.5 the Right Way
This one surprised me. The original poster says they only use Claude Opus 5.5, and on “medium” settings. Most people assume the highest effort setting is always best. In practice, medium gives you strong reasoning without slow answers or burning through your usage.
Here’s a simple rule I took from this:
- Medium: Your everyday default for writing, analysis and planning.
- Higher effort: Save it for hard problems like complex code, multi-step logic or big strategic decisions.
The expert says there are plenty more tips on this topic, so it’s worth reading their full take.
4. Take Your Privacy Seriously
The author’s reminder is direct: ChatGPT and Claude can read your chats. Many people paste client data, financials or personal details into AI tools without a second thought.
A few habits help a lot:
- Check your data and training settings in each tool you use.
- Remove names, account numbers and other identifying details before you paste anything in.
- Keep truly sensitive material out of consumer chat apps completely.
Use the tool all you want. Just know where your data goes.
5. Switch Between ChatGPT and Claude Without Losing Context
Plenty of people use both assistants, and losing all your context when you move between them is painful. The post’s author points out that you can switch without losing any context and data.
One easy approach is to ask your current assistant to write a briefing you can carry over. Try something like: “Summarize everything you know about my projects, preferences and writing style as a briefing I can paste into another assistant.” Save that summary and drop it into a Claude Project. You’ll be up and running in minutes instead of starting from zero.
6. Use Claude for Legal Work
The creator describes this as having “your own legal team.” I’d treat it as a very capable first pass, not as a replacement for a lawyer.
Where it really helps:
- Explaining contract clauses in plain English.
- Flagging unusual or one-sided terms before you sign.
- Drafting a list of questions to take to your attorney.
This saves time and money, and you’ll walk into legal conversations much better prepared.
7. Get Claude Certified
Here’s something many people miss. Anthropic offers official certificates, and this contributor highlights them as a path worth taking. A certificate gives you a structured way to learn, so you’re not just collecting random tips. It also gives you something concrete to show employers or clients, and that counts for more as AI skills become a standard part of job descriptions.
8. Stop Running Out of Limit
Hitting the usage limit halfway through a project is one of the most annoying things about Claude. The person who shared this says they have a set of hacks they use to never hit the limit.
A few habits that line up with that idea:
- Start a fresh chat for each new topic so Claude isn’t re-reading a huge thread.
- Edit your original prompt instead of stacking follow-up corrections.
- Keep shared files in a Project instead of re-uploading them over and over.
- Match the model and effort level to the task, as in point 3.
Twenty Minutes a Week Is Enough
The author ends with an honest admission: “I will not lie, you can’t master Claude in one day.” I appreciate that. No single trick turns you into an expert overnight.
Their suggestion is simple: set aside 20 minutes every week to stay in the loop. Small, steady learning beats one big cram session, especially with a tool that keeps changing this fast.
My advice is to pick one item from this list each week. Start with setup and prompting, since they make everything else work better. In two months you’ll be using Claude in a completely different way.
Check out the full LinkedIn post for the author’s complete list and all their tips on each topic.