The Magic First Message Formula
When you open up Claude or ChatGPT, your first message sets the entire trajectory. Here's the formula that works:
Start with the problem you're solving (not the solution)
Mention your future plans (this shapes technical decisions)
Be honest about your experience level
Ask if it's feasible and invite questions
Real Example That Worked
Here's an actual first message that led to a successful project:
I need a webapp that can store all my pet records in a logical way, like it organizes the PDFs and pictures of my pet medical records, and takes things like vaccinations and meds and conditions and displays them in an easy to read way. It's so I can show my petsitter and other healthcare providers. I need to make it so I can package into an iOS or Android app easily later. Assume I have no technical ability. I'd need complete step-by-step directions including where to code, how to start, the backend stuff, the authentication, the front end, anything else I'd need. I'm totally new. Is this feasible? Can you walk me through this? Ask questions if needed.
Notice what this does:
- • Clearly states the problem (scattered pet records)
- • Identifies users (pet sitters, vets)
- • Mentions mobile apps (influences tech choices)
- • Admits inexperience (gets beginner-friendly explanations)
- • Invites dialogue (not a one-way street)
Questions the LLM Should Ask You
The LLM will ask clarifying questions. If it doesn't, you might need to prompt it:
What else do you need to know to give me the best recommendation?
How can I avoid becoming a vibe code meme?
How do I tackle this without writing embarrassing AI slop?
What makes those components necessary?
Expect questions like:
- How many users will this have? (Just you vs. thousands changes everything)
- Do you have a budget for hosting? (Free tier vs. paid services)
- Do you prefer learning one technology deeply or keeping things simple?
- Will users need accounts or is it just for you?
- How comfortable are you with terminal/command line?
Red Flags in LLM Responses
Sometimes you need to push back or find a different approach:
Too Complex
If the LLM suggests Docker, Kubernetes, microservices, or "enterprise solutions" for your first project:
That seems overcomplicated for a beginner. Is that really necessary? What's a simpler approach? What makes those components necessary?
Too Vague
If you get high-level architecture talk without concrete steps:
Can you break this down into specific tools I need to install and exact commands to run?
Too Many Options
If presented with 5 different stacks:
If you had to pick ONE stack for a complete beginner that will still work when I have real users, what would it be and why?
Common Newbie Mistakes to Avoid
I want to build a way to share photos with my family
Why this matters:
Start small and specific. You can always expand later, but starting too ambitious guarantees failure.
I need to solve [specific problem], would AI help?
Why this matters:
Focus on solving problems, not using trendy tech. Let the problem dictate the solution.
I'm on a Windows laptop and need everything to be free initially. If things pick up, I'll be able to get a Macbook Pro
Why this matters:
Constraints shape recommendations. Be upfront about limitations to get realistic advice.
You suggested PostgreSQL - what is that and why do I need it?
Why this matters:
Never accept without understanding. It's your app - you need to know what each piece does.
Steering the Conversation
Once you get initial recommendations, keep the conversation productive:
For Technology Choices:
Why did you choose [technology] over [alternative]? What are the tradeoffs?
For Concepts You Don't Understand:
Can you explain [concept] like I'm someone who's never coded before? Maybe with an analogy?
When Things Get Too Technical:
I'm getting lost. Can we zoom out and explain what problem this solves before diving into how?
For Environment Setup:
I don't want to install a bunch of stuff on my laptop that I don't understand. Is there a cloud-based or sandboxed way to do this?
The Stack Discussion
Eventually, the LLM will recommend a "stack" — the collection of technologies your app will use. Here's how to evaluate if it's right for you:
Ask These Questions:
- 1. "How much will this cost per month at launch and at 1,000 users?"
- 2. "Which parts have generous free tiers?"
- 3. "If I need to switch [specific technology] later, how hard is that?"
- 4. "What's the learning curve like compared to alternatives?"
I know there are tradeoffs, but if YOU were building this as a solo developer who wants to launch quickly but not rebuild everything later, what would YOU choose?
Getting Actionable Next Steps
Before ending your brainstorming session, make sure you have:
The Shopping List
Can you give me a complete list of every account I need to create and tool I need to install before starting?
The Roadmap
What order should I build things in? What's the minimum viable version?
The Learning Path
What concepts do I need to understand before I start vibe coding with an LLM? What can I learn as I go?
The Folder Structure
Can you show me exactly how my project folders should be organized from the start?
The Follow-Up Sessions
Your first conversation is just the beginning. As you build, you'll have more conversations:
Session 2: Environment Setup
I'm ready to set up my development environment. I have [list what you prepared]. Walk me through the next steps. I need a tight JSON prompt to put in the project instructions so you have all the information needed to focus our efforts. Help me build this prompt and what files to keep in there for reference.
Session 3: First Code
My environment is set up. Help me build the simplest possible version that I can see in my browser and get a quick win.
Session 4: Problem Solving
I'm getting this error: [paste error]. I tried [what you tried]. What's wrong?
Switching Between LLMs
Different LLMs have different strengths. Don't hesitate to:
Claude Opus
Complex architecture discussions and code generation
ChatGPT
Debugging weird errors (it's seen everything)
GPT5
Brainstorming ideas
Claude Sonnet
Making HTML mockups to visualize UI
Gemini
Google Cloud-specific questions and graphics
Multiple LLMs
Cross-check important decisions
Your LLM Conversation Checklist
Before you start coding, make sure your LLM conversations have covered:
Remember
The LLM is your copilot, not your pilot. You're still making the decisions, but now you're making informed ones. Channel your inner Tech Program Manager.
If something doesn't feel right or seems too complex, speak up. There's always a simpler way when you're just starting out.
The goal of these conversations isn't to understand everything — it's to understand enough to start building. You'll learn the rest by doing.