You are probably using AI every day right now. You open ChatGPT, Claude, or Gemini in your browser, type a prompt, get an answer, and copy-paste it somewhere. That works for casual questions. But when you want to automate a repeatable workflow — like generating a CV for every job application — the browser becomes a bottleneck.
This is Part 1 of a 2-part series on building an AI agent with OpenCode. In Part 1, we will cover the fundamentals: what an AI coding agent is, why you need one instead of a browser, why OpenCode beats the alternatives, and how to write your first command. By the end, you will have a working AI agent that reads a file and returns a structured summary.
In Part 2, we will take everything you learned and build something real: a CV & Cover Letter Agent that reads your experience, analyzes a job description, rewrites your CV bullets with quantified impact, and generates a polished cover letter — all as a reusable command you can run again and again.
The ultimate goal of this series: help you move from copy-pasting prompts in a browser to building reusable AI agents that automate real workflows.
Let's be honest: most of us started with a browser.
You open ChatGPT, Claude, or Gemini. You type a prompt. You get an answer. You copy-paste it somewhere. You do this every day for writing emails, summarizing documents, brainstorming ideas, or debugging a small piece of code.
That works. But it has a hidden cost.
Every time you want to do something repeatable — tailoring a CV for a new job, rewriting a cover letter, summarizing a batch of files, or debugging a recurring error — you have to start from scratch. You re-type the same context, re-paste the same data, and re-explain the same instructions.
Here is what that looks like:
| What you want to do | In a browser (ChatGPT/Claude/Gemini) |
|---|---|
| Explain a function | Paste the code, wait for an answer |
| Refactor a module | Copy-paste, get suggestions, paste back |
| Run tests after changes | Manually open a terminal, type the command |
| Generate a CV for Job A, then Job B | Re-type everything each time |
| Debug an error | Paste the error message, hope for a fix |
| Automate a workflow | Not possible |
The problem is not the AI. The problem is the interface. You are using AI as a consultant — someone you ask questions of. But for real work, you want an agent — someone who does work on your behalf, in your project, with your tools.
That is where OpenCode comes in.
An AI coding agent is not just a chatbot. It is an AI system that can take actions in your development environment.
A chatbot answers questions. An agent does work.
Here is what an AI coding agent can do: