I don't see low-code and LLMs really competing with each other. It may be true for quick throwaway projects, but they solve different problems and can potentially be used together.
With low-code, users operate at a high abstraction level and get a deterministic output. You're fully in control of what the result will be. With LLMs you operate at an even higher abstraction level – just a prompt in plain English – but the output is non-deterministic.
So if you want fine control, you need to check line by line what it produced. I think it gets interesting if LLMs generate low-code instead of code. Users get the speed advantage of AI generation, but they can still understand and control what the software is doing.
It is equally important as the other low code solutions that have been floating around since the 80s. AI doesn't change the big picture, which is that when you empower non-technical people to create apps, they will do so. 9 out of 10 won't work well. Of the 10% that do, 9 out of 10 will be over-fit to the team that built it and not grow, even while solving a problem for that team. For the 1% that work, are broadly useful, and grow... they will fail at scale, and a professional team will then need to come in and smooth out the rough edges.
AI is a new tool to walk that same path. Maybe it will let people go father before needing help, maybe not. But if you are trying to run a low code platform, your focus should be at least partially on that last step of the path - how do you help people take their work farther before needing to call for help?
I think why people used low code, was because they were not too good with actual coding but still loved programming. So they used such tools which provide easy interface low code options for programming.
However, with the AI boom, and seeing what Claude can do, I also think that such people might now prefer to use Claude, depending on how complex application they are building.
Having said that, non-tech background would still prefer low code options, because not everyone can supervise an LLM.
I joined a few low code hackathons back in 2024, before this agentic stuff.
They're fully dead from what I see. With low code, you'd drag a component onto the screen, click it, look for a field (which may have a different name to the original field), fill that in, and then spend 30 min trying to align it on screen.
With agents, you just tell them what to do. Draw boxes on a piece of paper, take a photo with Claude on your phone, and you'll have a functioning UI.
If you wanted to modify layouts, you can do it straight from the toilet seat on your phone.
The other big feature in low code is maintaining API specs. You'd tell it what the tables are, what to connect to, data objects, and all that. Another thing that AI does better.
I suspect that people who use low-code prefer the node based editor (making UX assumptions here) will still prefer that UI when having to dig into what the Ai did. Similar to how developers (who are holding Ai right) will look at the code.
My other take on the title question, separate from your contextualization and based on my own low-code setup, is that having simpler abstraction layers that give more consistency in application implementation will be beneficial. It is very frustrating that they will implement the same thing in as many different ways as there are days in the week.
My context is low-code via text/template from a declarative config / data.
I don't see low-code and LLMs really competing with each other. It may be true for quick throwaway projects, but they solve different problems and can potentially be used together.
With low-code, users operate at a high abstraction level and get a deterministic output. You're fully in control of what the result will be. With LLMs you operate at an even higher abstraction level – just a prompt in plain English – but the output is non-deterministic.
So if you want fine control, you need to check line by line what it produced. I think it gets interesting if LLMs generate low-code instead of code. Users get the speed advantage of AI generation, but they can still understand and control what the software is doing.
What is your low-code service?
/!\ Disclaimer, I'm building in this space[0].
[0] https://breadboards.io
It is equally important as the other low code solutions that have been floating around since the 80s. AI doesn't change the big picture, which is that when you empower non-technical people to create apps, they will do so. 9 out of 10 won't work well. Of the 10% that do, 9 out of 10 will be over-fit to the team that built it and not grow, even while solving a problem for that team. For the 1% that work, are broadly useful, and grow... they will fail at scale, and a professional team will then need to come in and smooth out the rough edges.
AI is a new tool to walk that same path. Maybe it will let people go father before needing help, maybe not. But if you are trying to run a low code platform, your focus should be at least partially on that last step of the path - how do you help people take their work farther before needing to call for help?
I think why people used low code, was because they were not too good with actual coding but still loved programming. So they used such tools which provide easy interface low code options for programming.
However, with the AI boom, and seeing what Claude can do, I also think that such people might now prefer to use Claude, depending on how complex application they are building.
Having said that, non-tech background would still prefer low code options, because not everyone can supervise an LLM.
I joined a few low code hackathons back in 2024, before this agentic stuff.
They're fully dead from what I see. With low code, you'd drag a component onto the screen, click it, look for a field (which may have a different name to the original field), fill that in, and then spend 30 min trying to align it on screen.
With agents, you just tell them what to do. Draw boxes on a piece of paper, take a photo with Claude on your phone, and you'll have a functioning UI.
If you wanted to modify layouts, you can do it straight from the toilet seat on your phone.
The other big feature in low code is maintaining API specs. You'd tell it what the tables are, what to connect to, data objects, and all that. Another thing that AI does better.
I suspect that people who use low-code prefer the node based editor (making UX assumptions here) will still prefer that UI when having to dig into what the Ai did. Similar to how developers (who are holding Ai right) will look at the code.
My other take on the title question, separate from your contextualization and based on my own low-code setup, is that having simpler abstraction layers that give more consistency in application implementation will be beneficial. It is very frustrating that they will implement the same thing in as many different ways as there are days in the week.
My context is low-code via text/template from a declarative config / data.