How do I even realise the app have a bug? Is it like the customer has to find out by trying?
I have been working on this project, something that’s already there in the market. I am trying to bring in changes. I am trying to make a lot of things. I don’t know if I’m trying to reach a high branch and I am looping on this thought.
I tried contacting the company called GeekyPants. They are pretty good and within my budget oo. I have decided to work with them to get my product complete nad to launch them into market. Thanks to everyone who replied.
While vibe coding can help create many cool things, what you noticed is one reason developers are unlikely to disappear anytime soon.
Things like architecture, planning, UX/UI design, and testing may not always be the most exciting parts, but they are essential for turning an idea into something solid and reliable.
A demo is one thing; building and maintaining a real product is another. AI is very useful, especially for prototyping and speeding up development, but it still benefits a lot from someone who understands the underlying software engineering decisions.
So I would not say AI is useless at all. Quite the opposite: it can be incredibly powerful when used with the right knowledge and care. The stronger your fundamentals are, the better you can guide it toward building something that truly works.
The answer is fairly simple, but the work involved to close the gap might be large: implement thorough behavioural and unit testing. The testing framework depends on your architecture, but if web based you would normally consider playwright for behavioural tests.
You can also consider remote logging for production exceptions whereby each client would notify your server if it hits a caught exception. Make sure you write your code to use a standard exception system and architecture.
For the chosen programming language, C# or Python, Rust. Learn the ropes of that programming language, what are the patterns used in that particular language. Then learn about the target OS where you want to use/implement this idea. What are the requirements? What functionality can be done? what are no go’s. Learn and know your environment, even with vibe coding, it’s good to know what you are asking for.
This is a common product/engineering gap rather than a “user finds it” situation. In practice, apps shouldn’t rely on customers to detect bugs—proper observability (logging, error tracking like Sentry, analytics events, and crash reporting) should surface issues early. Even basic input validation + feature flags can help isolate changes safely. If you’re iterating fast (“vibe coding”), setting up a lightweight monitoring loop is what prevents regressions from reaching users in the first place.
I have an idea for a shopping app that will exclusively sell maternity wear. Is it a good idea to integrate an Ai chatbot in that?And I plan to hire an outsourcing team, is that practical?
If your app is very early and support volume is low, hiring a person can sometimes be cheaper, especially if you hire from developing countries, your support is in one language, and the working time is similar.
But this applies only in limited cases. Even if someone agrees to work for $10/day, that is still around $300/month for one person. And one person cannot handle 24/7 support, multiple languages, or the volume of questions at scale.
AI chatbots are heavily used in e-commerce because they can answer product questions, suggest products, handle FAQs, support different languages, and reduce first- and second-layer support work.
But the AI chatbot space is also very competitive. Many platforms claim they are the best. So always try it first with your product data and get quick customer feedback. If it works for your brand, then continue. Always check whether the platform allows testing properly before locking you in.
There are some good tools that you can try like Intercom, YourGPT, Tidio, or build your own using OpenAI Agent SDK if you have a bit technical understanding.
I’m not a true developer so my answers might be easier to understand.
(1) Ask the Coding AI to look for and fix bugs. That is vibe coding basics. It’s risk for developers who want to control the AI and don’t AIs just going off and making changes because they sometimes change what you don’t want changed, but for people who need the AI to be the primary coder (people like me), it works really well to have it search for and find it’s own bugs.
(2) You click through your app testing all the features using the Dev Tools panel on your browser to report back to the AI when you hit issues.
(3) Some bugs are harder for the AI to find and to fix. When you run into that, I recommend multiple AIs (getting a second opinion). This is a super skill so I recommend you practice it.
(4) Make sure the AI is noting the types of errors it makes most often. Could be your prompting, or the type of code/app, or something else in the context - doesn’t matter - make sure it documents its errors so in the future, it becomes less likely to repeat them.
This list will handle most of your issues but there are many more tips you find and learn.
Yes - it’s a good idea to integrate a chatbot into your app. And it’s not hard at all. You could ask Codex to do that for you. The hard part is WHAT you want the chatbot to do. Start easy, make it answer questions about the app (how to do X and Y). Then expand from there. Codex can easily figure out all the direcitons to use your app and then put that into the RAG (long term memory) for the chatbot. That’s probably 5-7 hours of work, if that. If I was going to hire anyone, I would do as much as I could myself first. Then they just need to check it but frankly, not many people will do better than Codex out of the box. The advantge of working with Codex first or only is that you can update your AI whenever you want. While it might not mean you ca have the world’s best chatbot, improvements in your chatbot will come as Codex improves. Altman said a couple of years ago ‘build on top of OpenAI’ so that you scale with OpenAI.
You don’t wait for customers to find bugs—that’s the expensive way to do it. Good engineering teams build layers to catch them early: unit tests, QA flows, staging environments, and real user feedback loops before full release. Bugs are expected, but unmanaged bugs are what hurt.
On the second part—what you’re feeling is actually common when you’re building something already existing in the market. You end up overthinking scope and direction. The key is narrowing down: what’s the one or two things you’re improving, and validating those quickly instead of expanding endlessly.
And it’s a solid move to work with an experienced team if it fits your budget. Just make sure you stay clear on outcomes, timelines, and what “done” looks like so you don’t get stuck in iteration loops.
How deep you understand the domain the app will be solving problems for (the deeper your understanding is the better for you but also may mean that standard builders like lovable etc will be limiting your exact needs)
How complex the domain and the problem is (usually complex problems require more custom, while easy flows or domains may fit into out of the box tooling)
