Production-ready Saas product with ChatGPT?

[SeedPhrase]

Can you build a complete Saas product that is production-ready just by using ChatGPT??(with little to no coding knowledge)

what do you think?

 

[AussieLocalSEOGuy]

Been wondering the same thing.
I know it can take something as simple as a paper sketch and transform that into a website.
And of course, generate code in multiple languages (but still needs guidance so some experience is necessary - I used it for VBA and fiddled around in javascript).

I heard that it can build chrome extensions. Haven't tried it, but that is probably the simplest time-to-market approach to validate any new SaaS.

 

[MAF Publishing]

Do you mean can chat GPT code that well? Or are you asking if GPT is ready to be the core of a production LLM app?

 

[HustleTong]

I don't think so
You'll still have to correct errors and guide the AI to avoid exploits in the code

 

[MAF Publishing]

It would depend the complexity of the service and your ability with GPT. But a really simple one that only performs one or two functions and has very minimal UI/account management, I think so. Once your code gets beyond the size you can easily manipulate within GPT's context window, you need real knowledge.

 

[Vikjohn]

Can you build a complete Saas product that is production-ready just by using ChatGPT??(with little to no coding knowledge)

what do you think?

No it can't, the thing, it can do simple stuffs but can't create a production ready aaas, if you have an experience with programming before, it would definitely be helpful, it would make your work alot more faster, but you can't use it it to make a production ready saas, when it Write codes it get to a point where it become delusional most especially when not writes alot of codes, it just forget things, create functions that don't even exist, you need to make most corrections yourself, chatgpt isn't mean to build stuffs, it's meant to assist you in doing somethings, it's like giving a calculator to a maths student.

 

[TMONEX]

I think it's possible if you can break your process into chunks of atleast 100steps.

People that want it to create production ready codes in one click are the problem. It has limits.

For example to create a perfect homepage with bootstrap.

You have to guide it a thousand times but it knows bootstrap framework so get it to start from there starting from creating your foundation, body, code structure etc.

In the end , you will be spending more time fighting with it. When you could have just writing the code yourself.

 

[MKSKS]

With little to no coding not possible.
But yes if you're a developer AI can make your life easier by saving time.
I use it a lot with logical headaches, SQL queries, database & code debugging & improvements.
Not recommended for code generation though.

 

[mindinSOCIAL]

Maybe it's possible with a little knowledge. At least you can imagine what you will print and manage and for what purpose. If Have zero knowledge, you'll be dancing around GPT lol

 

[AussieLocalSEOGuy]

I think it's possible if you can break your process into chunks of atleast 100steps.

Good idea. The less unfavorable approach is to chunk it up. Much easier also to build something as small as possible (MVP or smaller).

This approach could work. the analysis to strategic execution for code project part I've done with it was quite good.

Give gpt4 the business requirements or architecture.
Request to create a high level strategic plan of execution.
discuss this until satisfied.
take the high level plan and use it as the context for each of the main code sections. Put this plan on third party website if the context prompt primer can't handle it.
For each main section refer to third party website of the big plan (use scraper plugin or other way that grabs this page of text)
Get gpt4 to build each section just 1 at a time. (This reduces the memory max out where it falls back to gpt3.5).
repeat next until done.

Note between each section parameters may need defining and passing. This might require a separate planning task or integrated into the master high level plan.

 

[indianmojojojo]

Basic codes yes. SaaS functionality/ integrations no.

Plus you need to clean the code and tweak.

The upside is atleast 40-50% of the man hours spent coding goes down.

 

[gemini2023]

I would look at autogpt or using llangchain with LLM for making SaaS services. They can also interact with local LLM's