Browser Chat: Your AI Advisor
What you'll learn
~15 min- Have 3 real AI conversations and compare the results
- Identify what browser chat excels at (and what it cannot do)
- Write effective prompts for brainstorming, explaining, and reviewing
- Understand the 'advisor' mental model for browser-based AI
You will have 3 real conversations with AI and learn what makes browser-based AI tools powerful — before you ever touch the terminal.
Why this matters
Browser chat is the gateway. Most people start their AI journey here — opening ChatGPT, Claude, or Gemini in a browser and typing a question. That is completely normal and, for many tasks, it is the right tool.
Before we dive into CLI tools later in this module, we need to establish a clear picture of what browser chat does well and where it falls short. Mastering browser chat makes everything else easier. Once you understand its strengths and limits, you will know exactly when to reach for a more powerful tool. But first, let’s get good at this tool.
Mental model: AI as your advisor
Think of browser chat as a knowledgeable advisor you can consult anytime. Like a good consultant, it can:
- Brainstorm options and ideas with you
- Explain concepts in whatever way helps you understand
- Review your plans and point out what you might be missing
- Draft documents, emails, and proposals
Unlike a human consultant, it is available 24/7, it never judges your questions, and it responds in seconds. That makes it incredibly useful — as long as you understand what it is (and is not) capable of.
One important thing to know: the AI relies entirely on the context you provide. It does not know anything about your project, your background, or your goals unless you tell it. The more specific and detailed your prompt, the more useful the response will be.
Hands-on: 3 conversations right now
Open your browser and go to one of these tools. Pick whichever you already have an account for:
- ChatGPT (free tier available — check current pricing and limits)
- Claude (free tier available — check current pricing and limits)
- Gemini (free with Google account — check current pricing and limits)
You are going to paste 3 specific prompts and study the results. This is not hypothetical — actually do it.
If one tool is unavailable or you hit a usage limit, switch to another provider. If the free tier limit is reached, continue later or try an alternate tool. Any modern AI chat tool will work — the exact wording of responses may vary, but the patterns are the same.
Conversation 1: Ask for a project idea
Paste this prompt exactly:
I want to build a personal portfolio website. Give me a project plan with 5 steps. For each step, estimate the time for a complete beginner.
What to notice: The AI gives you a structured plan with time estimates. It adapts its language to “complete beginner.” This is brainstorming and planning — browser chat’s sweet spot.
Conversation 2: Ask for an explanation
Paste this prompt in a new conversation:
Explain the difference between frontend and backend like I’m planning a restaurant. Frontend is what diners see, backend is the kitchen.
What to notice: The AI picks up your analogy and runs with it. It maps technical concepts to something familiar. This is explaining and teaching — another area where browser chat shines.
Conversation 3: Ask for a review
If you are continuing in Conversation 1, the AI already has the plan in its context, so simply type:
Review this plan. What am I missing? What could go wrong?
If you are starting a new conversation, paste the plan from Conversation 1:
Here’s my plan for a portfolio site: [paste the plan from Conversation 1]. What am I missing? What could go wrong?
What to notice: The AI critiques the plan, identifies gaps, and suggests improvements. It catches things you might not think of. This is reviewing — the third superpower.
You used three different AI interaction patterns: generating (creating a plan), explaining (teaching a concept), and reviewing (critiquing work). These three patterns cover most of what browser chat is great at. Remember them.
What browser chat excels at
Based on what you just experienced, browser chat is the right tool when you need to:
| Task | Why browser chat works |
|---|---|
| Brainstorm ideas | It can generate many options quickly, helping you evaluate ideas faster |
| Explain concepts | It adapts explanations to your level and analogies |
| Review plans or writing | It catches gaps and suggests improvements |
| Write drafts | Emails, proposals, reports, documentation |
| Learn something new | You can ask follow-up questions in real time |
| Compare options | ”What are the pros and cons of X vs Y?” |
What browser chat CANNOT do
This is equally important. In a standard browser chat session:
- Cannot automatically access your local files. You can upload a document to some tools, but the AI cannot browse your computer’s folders or see your project structure on its own.
- Cannot run code in your local environment. It can write code for you. Some tools offer sandboxed code execution in the browser, but they cannot run your project’s actual backend or interact with your local setup.
- Cannot make changes to your project. It cannot create files on your computer, edit your project files, or install software.
- Cannot see your screen. It does not know what application you have open or what error you are looking at (unless you paste it in).
- Has a limited context window. In long conversations, the AI may lose track of earlier messages. If responses seem disconnected, start a new conversation and re-share the key details.
This is not a flaw — it is a design boundary. Browser chat is a conversation tool. It talks about things. Later in this module, you will learn about tools that do things. The opportunity is in learning to use both together.
When you paste code, documents, or project details into browser chat, that data is sent to the AI provider’s servers. Depending on the provider and your account settings, your input may be used for model training unless you opt out. Avoid pasting sensitive credentials, proprietary code, or regulated data without checking the provider’s data usage policy first.
When browser chat gives you code, you have to manually copy it, figure out which file it goes in, paste it, and hope it works. If it does not work, you copy the error back into the chat and repeat. This loop — copy, paste, fail, copy error, paste error, get new code, repeat — is exhausting and slow. There is a better way, and you will learn it in Lesson 3.
Try it yourself
No matter your field, browser chat adapts to your context. Here are examples for different domains:
🧬In Your Field: Biotechclick to expand
Browser chat for biotech: Ask it to explain a protocol, compare reagents, brainstorm experimental designs, or review a grant abstract. Example: “I’m designing a CRISPR experiment to knock out gene X in cell line Y. What controls should I include? What are the common pitfalls?”
📊In Your Field: MIS / Businessclick to expand
Browser chat for business: Ask it to draft a business case, compare analytics platforms, brainstorm KPIs (key performance indicators — measures of success) for a new initiative, or review a project proposal. Example: “I need to present quarterly sales data to the executive team. What visualizations tell the clearest story? What should I avoid?”
🏛️In Your Field: Government / State Devclick to expand
Browser chat for government dev: Ask it about compliance requirements, compare deployment strategies, brainstorm architecture options, or review a technical design document. Example: “We’re migrating a legacy .NET app to the cloud. What are the FedRAMP (a US government cloud security compliance program) implications? What questions should I ask the security team?”
Use AI output as a draft, not a final authority. Always verify compliance, legal, scientific, or regulatory claims with trusted sources. Do not paste sensitive or regulated data into browser chat without checking your organization’s policies.
When things go wrong
When Things Go Wrong
Use the Symptom → Evidence → Request pattern: describe what you see, paste the error, then ask for a fix.
Browser-based builders
So far, we have been talking about browser chat as an advisor — a tool you converse with. But a new category has emerged: browser-based builders that can actually create and edit code, not just talk about it.
- Claude Code web (
claude.ai/code) connects to your GitHub repos and can clone, edit, test, and commit code — no terminal required. Requires a Pro or Max subscription. - ChatGPT with code interpreter can run Python, generate files, and process data uploads — all in the browser. Limited to sandboxed single-session workflows.
- GitHub Codespaces gives you a full Linux terminal and VS Code editor in your browser, where you can install any CLI tool. Free tier available.
These are not fallbacks. For many tasks — especially if you cannot install software on your machine — they are legitimate building tools. The main tradeoffs are internet dependency, subscription costs, and (for some) less control over your local environment.
See Module 5, Lesson 4: Choosing Your Cloud Sandbox for a detailed comparison of every cloud sandbox option, including what each can and cannot do.
Key takeaways
- Browser chat is your AI advisor. It brainstorms, explains, reviews, and drafts — all through conversation.
- Three core patterns: generate ideas, explain concepts, review work. These cover most browser chat use cases.
- It cannot directly work with your files or local environment. That is what CLI tools are for — and you will learn about them next.
- The copy-paste loop is slow. Getting code from browser chat into your project takes effort. There is a better way coming in Lesson 3.
- Specificity matters. The more context you give, the better the response. Generic prompts get generic answers.
- Cloud sandboxes exist. Claude Code web, ChatGPT code interpreter, and Codespaces can build real things without a local terminal. See Lesson 4 for details.
- When you need AI to act on files, run commands, or automate workflows, that is when CLI tools become the better fit. You will explore that in the next lesson.
You want to understand what 'responsive design' means before building a website. What's the best approach?
Your browser chat gives you a Python script. What happens next?