I’ve Been Coding with AI Since GPT Came Out. Here’s What I Learned
When GPT‑3 First Blinked onto My IDE, I Wasn’t Sure What to Expect. Today, It’s Become My Most Relentless Coding Companion.
When GPT‑3 first arrived, I was intrigued. Suddenly, the walls between “idea” and “implementation” blurred: AI scaffolding replaced repetitive tasks, and conversational coding guidance filled knowledge gaps in real time.
By pairing GPT’s generative prowess with time‑tested software principles, I found myself not just writing better code but growing into a developer.
In those early experiments, AI became my tutor, my rubber‑duck debugger, and my co‑pilot all at once. And what started as an experiment quickly became the foundation of my craft.
True Power Unlocked
The First Spark: I asked GPT to create a Django view and accompanying template for a simple contact form. In seconds, I had a ContactView with form handling logic, URL route, and a ready‑to‑use HTML template with fields and a submission button.
Benefit: AI gets your core functionality up and running—wiring views, URLs, and basic front‑end—so you can immediately test user flows and tweak the UI.
Caution: Always step through the form logic yourself. Validate user input, handle errors gracefully, and adjust the template to fit your design and security requirements before going live.
Where AI Truly Excels
GPT shines brightest on repetitive or boilerplate tasks:
Project setup (Express routes, React components, CI pipelines)
Unit tests (Jest, Pytest, or whatever your team prefers)
Error triage (stack‑trace analysis or missing imports)
That means less time wrestling with syntax and more time architecting features that matter.
Where AI Still Stumbles
But don’t hand over your architectural vision just yet. GPT can’t:
Architect a microservices topology from scratch.
Read all your codebase and system specs.
Understand your team’s long‑term strategic goals.
Treat AI output like a pull request: review meticulously, write tests, and run your full suite before merging.
Why AI Falls Short
AI tools excel at generating snippets, components, or handling isolated tasks. But they struggle to connect these pieces into a scalable system. Unlike developers, they lack:
Contextual awareness: AI doesn’t understand the broader architecture, legacy code, or business strategy behind a project
Creative problem-solving: They remix existing patterns—they don’t invent new ones that suit your unique constraints or edge cases
Critical judgment: Making trade-offs between speed, security, performance, and maintainability requires experience and human situational reasoning—AI can’t do that.
As one Redditor put it:
“AI can provide single minded solutions … but architecting anything reasonably like an application buckles and you find yourself diagnosing issues for code you didn’t write.”
With Great Power Comes Great Responsibility
AI can make every developer feel like a “10x Dev”—and yes, it does remove much of the grunt work. But balance is key:
Learn First: Build a feature or fix a bug on your own to understand the fundamentals.
Leverage AI: Use GPT to explore optimisations, generate tests, or prototype alternatives.
Review & Refine: Critically evaluate AI output, adapt it to your codebase’s style and security posture.
Document Decisions: Keep clear change logs. Explain why you accepted or modified an AI suggestion.
This workflow minimises future bugs, preserves your hard‑won domain knowledge, and maintains code quality.
Bonus: Crafting a Winning Prompt
A well‑structured prompt is your secret weapon. Here’s one I recently used to jump‑start a client’s consulting website:
Prompt:
reate a professional and engaging website for a management consulting agency. The website should include the following pages: Homepage, About Us, Services, Case Studies, Blog, and Contact.Homepage: Feature a hero section with the agency’s name, a compelling tagline such as “Transforming Businesses Through Expert Consulting,” and a prominent call‑to‑action button like “Schedule a Consultation.” Include sections for a brief overview of services, client testimonials, and recent blog posts or news.
About Us: Provide information about the agency’s history, mission, values, and introduce the team with photos and short bios.
Services: Detail the consulting services offered, categorized if necessary, with descriptions and possibly icons or images.
Case Studies: Showcase successful projects with clients, including the challenge, solution, and results—preferably with metrics or quotes from clients.
Blog: Feature articles on industry insights, trends, and thought leadership pieces.
Contact: Include a contact form, the agency’s physical address, phone number, email, and links to social media profiles.
Design & Technical Requirements:
Use a color palette that conveys trust and professionalism (shades of blue and gray).
Choose a modern, readable font like Inter or Montserrat.
Ensure full responsiveness and mobile optimization.
Incorporate high‑quality industry images and subtle animations or transitions without distraction.
Provide intuitive navigation with a sticky header.
Optimize for search engines by including keywords like “management consulting,” “business strategy consulting,” and “organizational development” in content and meta tags.
Integrate a lead capture form or newsletter signup if possible.
Minimize load times by optimizing images and scripts.
Overall, the website should project expertise, reliability, and approachability, motivating potential clients to reach out.
Why This Works
Explicit Structure:
Listing each page and its required elements leaves no room for ambiguity. The AI knows exactly which sections to generate and what content each must contain.Concrete Details:
Providing taglines, button text, and content examples (testimonials, metrics, bios) gives the model specific anchors, leading to richer, on‑target output rather than generic placeholders.Holistic Guidance:
By covering visual design (colours, fonts, animations), UX (sticky header, navigation), and technical concerns (responsiveness, load times), the prompt ensures the AI delivers a well‑rounded, production‑ready scaffold.Business Alignment:
SEO keywords and lead‑capture instructions align the generated site with marketing goals, turning it from mere code into a strategic tool for traffic generation and client conversion.Flexibility for Creativity:
Phrases like “possibly icons or images” and “if possible, integrate” give the AI room to innovate while staying on brief, balancing prescriptiveness with creative freedom.Quality & Security Cues:
Implicit performance and optimisation requirements nudge the AI toward best practices—image compression, minimal scripts, and semantic markup—resulting in cleaner, faster, and more maintainable code.
By combining clear deliverables, concrete examples, design standards, and strategic objectives, this prompt turns GPT into a comprehensive partner, accelerating initial development while preserving control over quality and brand alignment.
Let’s Connect!
🐦 X/Twitter : alex_kazos
💼 LinkedIn: alex-kazos
👨🏻💻 GitHub: alex-kazos