I’ll admit something. For a long time, every AI conversation I overheard made me feel like I’d missed a class. Agents, chain of thought, connectors, effort levels. It sounded like a lot of studying just to get started.
Then I found a post from an AI professional on LinkedIn that clears all of that up. The author’s main point is “AI is not that hard.” After reading their breakdown, I agree. If you’re starting from zero, this guide walks through what they shared, one simple step at a time.
🧠 Step One: Stop Worrying About “Agents”
The author starts by asking you to forget the word “agents” for now. An agent is just an AI that can take actions for you, not only chat. The good news is you’ve probably been using one already without knowing it.
The original poster points out that these are all agents:
- Claude Code and Claude Cowork from Anthropic
- ChatGPT Codex and ChatGPT’s work tools from OpenAI
In the author’s words, “the technology just became invisible.” They compare it to something called “chain of thought,” or CoT. That’s a technique where the AI thinks through a problem step by step before it answers. Most people have never heard of it. But CoT is what gave us “reasoning models,” which are the smarter AIs we all use now.
You don’t need to understand the engine to drive the car. You just get a smarter AI, whether or not you know how it works inside.
I found this really freeing. You can skip the technical vocabulary and still get great results.
🤖 Step Two: Pick the Right AI
With that out of the way, the next question is which AI to use. The expert explains two simple ideas:
- Models are different versions of an AI, and each has a different level of intelligence.
- Effort levels control how hard the AI “thinks” before answering. More effort usually means a smarter but slower answer.
Here are the creator’s recommendations as of Tuesday, October 13th, 2026:
- Use Claude Opus 5.5 on medium effort for most things.
- Switch to high effort when it’s a very big task.
- Use the Claude app, because the author says it works better there.
- ChatGPT is just as good, also in the app. If you go that route, use Sol 6.1 on medium.
- Gemini, Copilot, Grok and others are fine, but they aren’t as well integrated and don’t have as many good features yet.
For a beginner, this is great because it removes the guessing. Pick one app, set it to medium, and only turn it up when the job gets big.
✍️ Step Three: Write a Simple Prompt
Once you know which AI to use, the next step is talking to it. A “prompt” is just the message you type into the AI. The post’s author says people overthink prompts the same way they overthink agents.
According to the author, a good prompt only needs three things:
- A goal: what exactly do you need to achieve?
- A context: what should the AI know before doing the task?
- A closing statement: “ask me questions if needed”
Here’s the template the expert shared. You can copy it directly:
Prompt of the Day
1 – I want [goal] so that [success].
2 – My situation is [context].
3 – Ask me questions until you have enough context to start.
That last line does a lot of work. Instead of guessing what you mean, the AI asks you follow-up questions first. You end up with a much better answer without having to write a long, perfect prompt.
Example for a lawyer:
I want to negotiate a lease exit if permits fall through so that my client isn’t paying rent on an unusable space. I represent a restaurant taking over a former bank. Ask me questions until you have enough context to start.
Example for a salesperson:
I want to help our champion defend our price to procurement so that we keep our margin. We sell warehouse software & procurement wants 20% off before signing. Ask me questions until you have enough context to start.
The creator sums it up simply: goal, plus context, plus “ask me questions.” Their reaction is “bravo, you know AI.” Honestly, I think this formula alone will improve most people’s results right away.
🔌 Step Four: Connect AI to Your Apps
The original poster admits they got a little too excited and that there’s a bit more to it. The next level is connecting AI to the apps you already use every day, like Gmail, Outlook and Slack.
When an AI is connected, it can read and work with your real information. The author says they don’t draft a single email without Claude anymore. Here’s where to find this feature:
- In Claude, they’re called “connectors”, and you’ll find them inside the + button.
- In ChatGPT, they’re called “plugins”, in the left menu.
The author’s favorite connectors are Wispr Flow, Granola and Gmail. For coding work, they also use GitHub, Vercel, Supabase, Resend and Apify.
A practical tip for beginners: start with one connector, such as your email. Ask the AI to draft a reply to a real message, then edit it yourself. You’ll quickly see why the author relies on it.
💻 Step Five: “Code” Without Coding
This is what the author calls “level 3.” They know that tools like Claude Code or Codex can feel intimidating. Still, they make a fun point: everything is code. An image is code, a PDF is code, an Excel file is code. They even joke, “you are code?”
Their view is that coding is pretty much solved, so you might as well use it. And you don’t need to learn programming to do it. The approach is the same as before:
- Use the same goal + context + questions prompt.
- Connect GitHub and Vercel. The author says “trust me” on this one.
- Ask for ambitious goals.
For context, GitHub is where code gets stored, and Vercel puts websites and apps online. Together, they let the AI build something real and publish it for you. You describe what you want in plain words, and the AI writes the code.
📌 The Honest Ending
The post wraps up with a funny twist. The author says they just hit the 3,000-word limit and adds, “I guess mastering AI is not that simple.” I love that ending because it’s true. The basics really are simple, but there’s always more to explore once you’re comfortable.
If you’re just starting, here’s your plan from the creator’s advice: pick Claude or ChatGPT, set it to medium, use the three-part prompt, then add one connector. That’s a solid foundation.
Check out the full LinkedIn post to read the author’s original breakdown in their own words.