Can CRUD softwares survive in this AI era? | Hacker News Reader

Can CRUD softwares survive in this AI era?

4 pointsDheerajiitr8 comments

Had a funny incident recently:

I created a clone of a SaaS product and shared it with the founder and he was pissed (scale, security etc etc), This led me to think, is there any product moat left in SaaS (I know distribution is there).

With time do you'll think that every department can create AI enabled internal softwares needed for their operations ?

Do you think organization's would want to shift from per user pricing model to a fixed maintenance cost for their in house softwares ?

View on HN

I think traditional CRUD SaaS software will have more usage than ever. Between LLM function calling and small niche SaaS requirements, SaaS market will continue to grow.

But I think VC funding for them has mostly dried up because the moat is mostly gone given how much easier it will be to create SaaS.

Thus, I believe we will have smaller SaaS companies, more total users overall, and an overall bigger SaaS market.

I think it's similar to what is happening in the server CPU market. All the attention is given to Nvidia GPUs. No one really cares about server CPUs anymore. I think server CPUs will make a huge comeback because guess what the services that LLM function calling will run on? CPUs. The demand for CPUs will increase with AI, not decrease. LLMs + tool use is the model I expect in the next 1-5 years. I don't think LLMs will do it all. There is an old saying in software development: use the best tool for the job. Well, LLM inferencing won't always be the best tool for the job and LLMs should be smart enough to figure out which tool to use instead.

Kind of agree with what you are saying, though what I am thinking is "Does it even make sense for companies to spend $20/emp/mo for CRUD based softwares" ?

Would you continue paying for products like ERP, HRMS, ITSM, Jira etc etc ?

I think we'll start to see more $5/month for a SaaS because it costs a lot less to create one. We're going to have more micro SaaS that meets the needs of the niche.

I think the $20 ones will have their place for businesses and enterprises.

Maybe SaaS APIs need some kind of standard, usage-based billing method to cope with a million different AI services calling all different APIs. It would be much easier to integrate loads of services at 0.08c per call through some common payment provider without needing to go to each one manually, set up billing, pay $20/month, generate a key, etc.

I think CRUD API may remains (called by chatbot), but CRUD app may not.

I got this insight making a shopping mall chatbot from a backend server. It has 289 API functions, and all of them are properly called in the AI agent.

With this experience, I think that many applications may be transformed to chatbot, and only specific purposes apps will be survived.

Agreed, the API's are must have and agents can perform actions on our behalf.

I don't think every form UI to CRUD API becomes a chatbot interface. First and foremost is the validation of data that leads to chatbot asking multiple questions to resolve data consistency and correctness. Another way to look at is, with CRUD data, chatbot can fill the form UI if it passes all validation & asking user nod before it updates the system. Not all CRUD will transform to chatbot immediately but they may be augmented by chatbots to make the whole interaction as automated and easier, they'll co-exist until AGI can handle every situation perfect

Great insight making me think a lot

Reply on news.ycombinator.com