Posted on 02/23/2026 2:52:04 PM PST by aquila48
ZDNET's key takeaways
- Treat the AI like another developer, not a magic box.
- Encode design systems and user profiles in system prompts.
- Every fixed bug becomes a permanent lesson learned in the project's DNA.
Ever since the days of punched cards, I've self-identified as a programmer and a computer scientist. The programmer side is the practical side of my engineering identity, the person who crafts code line by line. The computer scientist is the theoretician, the scientist, the strategist, and the planner.
While I love the theory and science of computers, I've always enjoyed the hands-on feeling of cutting code. I think it's probably akin to how some woodworkers prefer hand tools over power tools for the visceral feel of working with wood.
Also: Worried about AI coding? Why the invention of power tools is the blueprint for your career future
Unfortunately, I've never had much time to code. My day-to-day job has been as a company executive, founder, educator, and writer. I do love making software products, but I've never managed to get more than one small product done each year, using little bits of available nights and weekend time.
All that changed this past September. That's when I started using agentic vibe coding tools, such as OpenAI's Codex and Claude Code.
Since September, I've built and shipped four major products (WordPress security add-ons), built a working iPhone app for managing 3D printer filament, and am close to having a beta of an app my wife requested for managing sewing patterns. These last two are being built simultaneously for iPhone, iPad, Apple Watch, and Mac.
As a sole coder, agentic AI has been a force multiplier of almost breathtaking capability.
Also: I got 4 years of product development done in 4 days for $200, and I'm still stunned
In this article, I'm going to take you through seven best practices I use. These practices help me work with AI as a partner, and generate products of a quality suitable for production use. At the end, I'll also share a bonus best practice that comes in handy more often than you might expect.
This is vibe coding. But it's vibe coding with engineering discipline, and an underlying framework designed for robustness and product quality. If you want to use AI to build your apps, follow these best practices.
Primary practice: Written instructions The items listed below are specific, deliberate practices. Each one comes from something I purposely built into my workflow.
The way I make these practices stick is I've added them to the "ini" files for the AIs, the CLAUDE.MD and AGENTS.MD files. I've also added other files used to document the project itself. I'll describe those in more depth as you read the rest of this article.
Let's start with my first best practice, codified when I found that agent behavior in Xcode was unreliable for multiple parallel processes.
Definitely keep reading until the end, because the aforementioned bonus best practice can be a real game-changer.
1. Sequential visibility over parallel speed
......
Dear FRiends,
We need your continuing support to keep FR funded. Your donations are our sole source of funding. No sugar daddies, no advertisers, no paid memberships, no commercial sales, no gimmicks, no tax subsidies. No spam, no pop-ups, no ad trackers.
If you enjoy using FR and agree it's a worthwhile endeavor, please consider making a contribution today:
Click here: to donate by Credit Card
Or here: to donate by PayPal
Or by mail to: Free Republic, LLC - PO Box 9771 - Fresno, CA 93794
Thank you very much and God bless you,
Jim
"As a sole coder, agentic AI has been a force multiplier of almost breathtaking capability.
Also: I got 4 years of product development done in 4 days for $200, and I'm still stunned"
If you're in programming, better think of switching to plumber. Pays better and the job will be around longer, at least another 10 years.
I retired and I still get job offers. AI can’t read the minds of the users. I can.
“AI can’t read the minds of the users. I can.”
That’s true, at least for now. Short term the jobs that will be in demand will for people who can translate users needs for a particular application into detailed and precise specs for AI.
It’s kind of a translator of high level marketing specs to more detailed and concise specifications for the AI and iterating the process till you get what you want.
And that process moves fast and in the process AI learns a lot about what humans want. So it will eventually be enough to feed it marketing specs for a product.
I have had the exact same experience
Regenerated what had been an 18 month PhD thesis project for a very bright PhD student in a little over a week
Stunning
I’ve been experimenting with Claude and Grok for various tasks and find it very useful - the software developer jobs at risk are among those less adept at using AI - for now. Perhaps later on it will lead to my early retirement.
built a working iPhone app for managing 3D printer filament, and am close to having a beta of an app my wife requested for managing sewing patterns. These last two are being built simultaneously for iPhone, iPad, Apple Watch, and Mac. I’m building one on how to put my wife in the olds folk home.
Basically the job will be Subject Matter Experts who use AI to generate the code. But they will still have to be knowledgeable. But the “Code” will be the Product Requirements Document.
Flutter or Swift? What LLM do you use?
Now my truck know how long I have to wait before I run the red light.
“Perhaps later on it will lead to my early retirement.”
A friend of mine just did that. He saw the writing on the wall... made sure he was set financially and then pulled the plug.
Claud and ChatGPT are amazing using it via VSCODE. Right now ChatGPT codex-max 5.2 is killing it!
When it screws up Claude fills the gap. ANd vice versa. But you get so much more bang for your buck on Codex max.
I’m volunteering to teach kids at an alternative school financial literacy and in particular how to buy a house.
2 of the sessions I had to find a presentation. Didn’t find what I wanted .
A prompt to Chat GPT got me exactly what I wanted in 45 seconds.
Saved hours of work.
I hike a lot and track all my hikes including distance, time, elevation gain, calories expended. I use MayMyRun and GaiaGPS to do the tracking. I’ve noticed wildly varying calorie estimates depending how many stops I make for the dog to poop, to sniff, for me to take photos, to stop and enjoy the view, etc. The tracker seems to take my maximum expenditure rate and apply it to the entire logged time irrespective of the stops.
For a while, I fed key parameters to Grok and had it do the calculations but it was time consuming to set up the chat conversation. I made a template, but moving data into Grok via that template was cumbersome and it was hard to make Grok aware of the grades and surface conditions.
Then, just last week, it dawned on me that I can export the GPS data from MapMyRun and build an app in Python to do all the calcs. I started using agentic coding with ChatGPT and now have an excellent 3,000 line program that parses the TCX file and applies three different metabolic exercise models for me. I keep enhancing the model and making changes, but right now it is outstanding.
This agentic coding absolutely blows me away. It would have taken me a year to accomplish what I’ve been able to do in ten days of part-time effort.
In computer programming, vibe coding is an AI-assisted software development practice. It is a chatbot-based approach to creating software where the developer describes a project or task to a large language model (LLM), which generates source code based on the prompt. The term was introduced by computer scientist Andrej Karpathy, a co-founder of OpenAI and former AI leader at Tesla, in February 2025. According to Karpathy, vibe coding typically involves accepting AI-generated code without closely reviewing its internal structure, instead relying on results and follow-up prompts to guide changes.I've paid attention to the architecture of the code I've gotten from ChatGPT. There are definitely shortcomings. One function is way too long and complicated and needs to be factored. ChatGPT will stick some formulas into long report printing sections.The term was introduced by computer scientist Andrej Karpathy, a co-founder of OpenAI and former AI leader at Tesla, in February 2025. Merriam-Webster listed it as a "slang & trending" term in March 2025. It was named the Collins English Dictionary Word of the Year for 2025.
Advocates of vibe coding say that it allows even amateur programmers to produce software without the extensive training and skills required for software engineering. Critics point out a lack of accountability, maintainability, and the increased risk of introducing security vulnerabilities in the resulting software.
When I call these out, it replies "You are right" and fixes them. Since I didn't architect the program from scratch, I'm not intimately familiar with it and its structure. When it says "replace these lines with this new code" it is not always easy to find the lines to be replaced, so I have to ask for context.
But, even with those issues, it's incredibly fast.
When a conversation gets long, ChatGPT slows way down (the "context window" got too large). Then I instruct it to start a new chat and tackle a new task. That speeds it up again for a while, but then the same slow-down happens after a couple hours.
But, all-in-all, this is amazing.
You have to be explicit and tell the coding style you want. I have Claude code process a document with best Python coding practices and tell it explicitly to follow the standards.
“This agentic coding absolutely blows me away. It would have taken me a year to accomplish what I’ve been able to do in ten days of part-time effort.”
And all the while it is learning from that experience!
It’s the documentation that really saves me a lot of time, I can input a couple of informal requirements docs and some tech specs, and get a full-blown system design from it.
“You guys start coding and I’ll go find what they want” is most specs
“You guys start coding and I’ll go find what they want” is most specs
LOL, so true.
Coding is easy, Debugging is hard. And also making sure the code is secure.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.