A friend called me last week. She's tech-adjacent and has heard nonstop about AI for two years straight. She wanted to know it if was too late, had the train left the station?

I've been thinking a lot about this and short answer: no, but it's time to get moving and make sure you don't get left behind.

I should say upfront: there are a million ways to approach this and I encourage everyone to figure out what works for them. However, there's been a lot of "do something with AI, do it now" and not a ton of "here's exactly what I would do if I didn't know anything about LLM's and wanted to start today."

I'm going to share what's working for me right now, which is mostly Claude and a few other tools that fit around it. The specific tools matter less than the fact that you actually start. Don't get stuck thinking about the "right" approach. Just pick something and go.

Step One: Get Claude and Wispr Flow

Go to claude.ai and sign up for the Pro plan. You want Claude Opus, which is Anthropic's smartest model. When Opus came out it changed the game for software developers and technical folks - this blog post isn't for them - it's for regular people like you and me.

Then download Wispr Flow on your computer or phone. Why Wispr Flow? In my experience, the biggest barrier to using AI isn't the AI - it's the typing and trying to figure out the right formatting, the right words, etc. Most people sit in front of a blank chat window and freeze, they don't know what to do: they end up overthinking the whole thing.

Wispr Flow is voice-to-text that actually works. You talk, it transcribes, and the output lands clean, no filler words, no broken sentences. You don't need to think about formatting or grammar. You just talk.

Step Two: Speak Like You're Talking to a Friend

This is the part people get wrong. They treat AI like a search engine and end up typing the same way they use Google - short keywords and questions that get generic answers. This approach causes a lot of people to abandon and say "I'll stick with what works for me today."

Try this instead, talk to Claude the way you'd explain a problem to a friend who actually cares. A friend who's smart, who has context, and who isn't going to judge you for rambling. Give it the whole situation. Don't filter yourself. Don't worry about how long it is or whether you're being articulate enough - just dump it all out.

"Hey Claude, I've been at my job for three years and I just got passed over for a promotion. My manager says I need to be more visible but I don't even know what that means. I do good work, I show up everyday but I don't know what to do differenty. Help me figure out what to do here - where should I start breaking this whole thing down?"

The above is an easy way to get started - Claude calls it a prompt. No special syntax, no magic words or fancy technical words, just a person explaining their situation to something that can actually help them think through it.

The combination of Wispr Flow and Claude is what makes this work. You're not staring at a blank text box trying to craft the perfect question. You're just talking out loud, the way you'd vent to a coworker or call a friend on the way home from work. The AI gets all the context it needs because you're giving it naturally, the way we actually communicate.

One thing I keep recommending to people who don't know where to start: ask Claude what it would do if it were in your situation. Just explain where you're at, give it as much detail as you're comfortable with, and say "what would you do?" It won't just give you a list of options. It'll walk you through how to think about it. It's a really good guide, and it's a much more natural starting point than trying to figure out the "right way to use AI."

Step Three: Set Up Projects

Once you've gotten comfortable just chatting with Claude, the next thing to learn is Projects. This is where it starts to get really powerful.

A Claude Project is basically a workspace with persistent context. You give it background information, files, instructions, and then every conversation inside that project has all of that context built in. You don't have to re-explain yourself every time you start a new chat.

Think of it like this: the regular chat is calling a smart stranger for advice. A Project is having a colleague who already knows your situation, your goals, your constraints, and your preferences.

I have Projects for different areas of my work. One for product strategy where, in the past, I've loaded in roadmap docs, competitive research, and customer feedback. One for writing where it knows my voice and style. One for meeting prep where it has context on the people I'm meeting with and what we've discussed before.

The setup takes maybe ten minutes. You drop in the relevant files, write a short description of what this project is for, and you're done. Every conversation after that is dramatically more useful because you're not starting from zero.

Step Four: Skills, Starting with Meeting Notes

Here's where people really start to feel the difference in their day-to-day. Claude has a feature called Skills, which are basically specialized instructions that tell it exactly how to handle specific types of work.

The one I'd start with: meeting notes and transcription.

Here's my exact workflow. I have a Pixel Phone, which comes with Google Recorder built in. It has automatic speaker recognition and the transcription is genuinely excellent. I hit record at the start of a meeting and forget about it. After the meeting, I open the web version of Google Recorder, clean up the speaker labels so it knows who said what, and export the transcript.

Then I drop that transcript into Claude with my meeting notes skill. What comes back isn't just a summary. It's structured notes with action items, key decisions, who said what, and the important stuff pulled out of a 45-minute conversation where half of it was small talk and tangents. Perfectly written, in my style, every single time.

I used to spend 10-15 minutes after every meeting writing up summaries, open questions, and action items. Now I spend about 30 seconds reviewing what Claude produces. And the output is better than what I was writing by hand, because it catches things I missed or forgot.

This is the moment where it clicked for me in a small but genuinely helpful way. Not the abstract "AI is going to change everything" stuff. The concrete "I just got 20 minutes back in my already jampacked day, now I can focus on higher level work that actually needs my effort and skill."

You Don't Need to Master It. You Need to Know What It Can Do.

Here's what I told my friend at the end of the call: you don't need to become an expert at this today. You don't need to understand prompting frameworks or AI architecture or any of that. What you need is exposure. You need to push on it enough to build a mental map of what it's capable of, where it falls short, and how it fits into the way you already work.

Try to break it. Ask it something you think is too complicated. Give it a task you assume it can't handle. See what happens. Sometimes it surprises you and sometimes it doesn't. Both outcomes are useful, because now you know.

Closing Thoughts.

I've been writing about work stuff and daily tasks, but I want to be real about something. The conversations I've had with Claude that surprised me the most weren't about meeting notes or product strategy. They were about the things that keep me up at night.

The stuff you'd normally only talk about with an expensive therapist because they're trained to pull it out of you. The decisions you're overthinking at 3 AM. The relationship dynamics you can't quite articulate. The career anxiety you don't want to admit to anyone because it makes you feel like you're behind. The life questions where you're not sure if you're doing it right and you don't want to be judged for asking.

I'm not saying Claude is a replacement for therapy. It's not. But there's a version of this that goes beyond "help me write an email." There's a version where you explain something complicated and personal, and what comes back makes you feel genuinely understood. Where the conversation is so thoughtful that you walk away from it the way you walk away from a really good talk with someone you respect, thinking "wow, that was actually helpful."

Once you have that experience, the whole framing of "should I be using AI" stops feeling like a tech question. It just becomes part of how you think things through.

If you really commit to making Claude one of the tools you use to get things done, think better, work faster, or just handle the stuff you don't want to deal with, the shift happens on its own. You don't have to force it. You just have to start, and then pay attention.

You're never too late. But today is a really good day to begin.