LV
4
So before we begin I should make one thing very clear:
I do not believe that AI can currently be used to build a 100% working product that can be released to the public.
With that being said, I do believe that it can be utilized to validate ideas and create early stage prototypes which can then be turned into MVP builds and beyond.
One of my favorite tools for doing this is Bolt - https://bolt.new/
I should also mention that I am not a coder and only hold a general understanding of the technical development process.
Now, Bolt.new is interesting, as a user you are presented with essentially a chat GPT prompt box that most people should be familiar with by now.
To begin the process it is literally as simple as entering a prompt and telling Bolt what you want to build.
What I do recommend for this first prompt is to include as much detail as possible while even providing links to other projects that you want to replicate the design or UI/UX of. It is highly recommended to spend some time building a well thought-out and thorough first prompt preferably using chat GPT or another AI to assist crafting the perfect first prompt for bolt to kick off your project.
Once you begin the process you will see that bolt begins building your project right in front of you and even provides a preview of what has been built so far so you can see each build update in real time and guide or correct the process as it takes place.
As long as your first prompt is extremely solid I would recommend starting with your big chat gpt prompt and then follow up with a prompt that says:
Continue building missing functions and features.
Bolt is not going to build your entire project in one shot, it is going to do it in pieces and charge you credit each time to make execute a prompt.
ERRORS
Dealing with Errors is one of the biggest complaints you will come across from Bolt users.
You can easily get stuck in unfixable error loops that drain your credits and leave you completely frustrated.
When this happens I would recommend attempting to fix the error a couple of times, before resorting to undoing the last prompt and going back a step. Sometimes you can fix an error loop by simply having it rebuild the last step again.
What you should NOT do is get stuck in an error loop, attempt to fix it multiple times (5 or 6) and expect that it will eventually snap out of it and you will recover your working prototype, it's much better to go back a step.
Bolt is not good at implementing design and it is not good at doing small ui and ux changes (sometimes it can do these things fix no problem, other times it just can't seem to understand even with screenshots, arrows, examples etc.
This is why I say it is good to use Bolt for prototype building and then bring in a real coder at some point to either complete the integrations or (my preferred way) have him rebuild everything based off your prototype concept.
AI will soon be able to build anything and you will have little old grandmas making Hollywood blockbusters and multi million dollar SaaS companies, but it's not quite there yet..... Check back in 6 months.
I do not believe that AI can currently be used to build a 100% working product that can be released to the public.
With that being said, I do believe that it can be utilized to validate ideas and create early stage prototypes which can then be turned into MVP builds and beyond.
One of my favorite tools for doing this is Bolt - https://bolt.new/
I should also mention that I am not a coder and only hold a general understanding of the technical development process.
Now, Bolt.new is interesting, as a user you are presented with essentially a chat GPT prompt box that most people should be familiar with by now.

Getting Started
To begin the process it is literally as simple as entering a prompt and telling Bolt what you want to build.
What I do recommend for this first prompt is to include as much detail as possible while even providing links to other projects that you want to replicate the design or UI/UX of. It is highly recommended to spend some time building a well thought-out and thorough first prompt preferably using chat GPT or another AI to assist crafting the perfect first prompt for bolt to kick off your project.
Once you begin the process you will see that bolt begins building your project right in front of you and even provides a preview of what has been built so far so you can see each build update in real time and guide or correct the process as it takes place.
As long as your first prompt is extremely solid I would recommend starting with your big chat gpt prompt and then follow up with a prompt that says:
Continue building missing functions and features.
Bolt is not going to build your entire project in one shot, it is going to do it in pieces and charge you credit each time to make execute a prompt.
ERRORS
Dealing with Errors is one of the biggest complaints you will come across from Bolt users.

You can easily get stuck in unfixable error loops that drain your credits and leave you completely frustrated.
When this happens I would recommend attempting to fix the error a couple of times, before resorting to undoing the last prompt and going back a step. Sometimes you can fix an error loop by simply having it rebuild the last step again.
What you should NOT do is get stuck in an error loop, attempt to fix it multiple times (5 or 6) and expect that it will eventually snap out of it and you will recover your working prototype, it's much better to go back a step.
Bolt is not good at implementing design and it is not good at doing small ui and ux changes (sometimes it can do these things fix no problem, other times it just can't seem to understand even with screenshots, arrows, examples etc.
This is why I say it is good to use Bolt for prototype building and then bring in a real coder at some point to either complete the integrations or (my preferred way) have him rebuild everything based off your prototype concept.
AI will soon be able to build anything and you will have little old grandmas making Hollywood blockbusters and multi million dollar SaaS companies, but it's not quite there yet..... Check back in 6 months.
Last edited: