The problem with this is that agents are neither going to be as reliable or as cheap to run as the software they produce that does it deterministically. I see what the author is getting at, that agents are getting more reliable, but I still think using the agent to write a tool is generally going to be a superior solution to asking the agent to do the work itself in every instance.
> agents are neither going to be as reliable or as cheap to run as the software they produce that does it deterministically
There are classes of software for which that is both unambiguously true, and will always be true. Handling HTTP requests, for example. I just find the trade-off has shifted from "you absolutely need to delegate this or program it" to also including "or just have a folder and get the agent to do it", and I'm finding this pattern pop up more and more for me.
Using ChatGPT or other services to convert the data between various formats indeed saves a lot of time and removes the need to write throwaway code to do that conversion. One should still carefully review the result, because the bigger the size of the data, the more chance of hallucinated results, even on the most recent models; but in general, this is a great improvement to some of my regular data conversion chores.
People need to stop pretending text and voice commands are an evolution on the user interface.
The UI is the innovative thing that simplifies and enables focus - not talking to “an agent”.
Imagine saying or typing what VIN you want instead of a dropdown menu with a list of them - isn’t it way faster and easier to see options and click one in a UI? Rather than have a back and forth conversation?
The take is really - if you zoom out - that we can cram all UX into a chat. Idk about that, I like a UI.
I’m more interested to see what kinds of UI/UX beyond chat we get. If you’ve messed with the AI video SaaS that’s already out there a few UX paradigms are already being established, like the concept of image references - they all do it different ways, but there will be an established way. Like pull to refresh or pinch zoom became established UX for smartphones.
That sounds like a property that's true of good UIs, rather than most UIs, and in particular most SaaS UIs or intra-company UIs. ClickUp's UI is bad enough that I would much rather not use it in preference to chat.
I'd say the inverse - it's only not true when it's a bad UI (like ClickUp's).
I agree though, sometimes text input is a superior UX than using any UI - like Spotlight search or git CLI.
Also agree that AI will broaden that - but the default?
A good thing to keep an eye on is image editing: Will people keep telling a chatbot what changes to make, like a slot machine hoping for the best, or will companies start putting more tools into the workspace (e.g. draw a selection box around an area and "Generative replace...")
The problem with this is that agents are neither going to be as reliable or as cheap to run as the software they produce that does it deterministically. I see what the author is getting at, that agents are getting more reliable, but I still think using the agent to write a tool is generally going to be a superior solution to asking the agent to do the work itself in every instance.
> agents are neither going to be as reliable or as cheap to run as the software they produce that does it deterministically
There are classes of software for which that is both unambiguously true, and will always be true. Handling HTTP requests, for example. I just find the trade-off has shifted from "you absolutely need to delegate this or program it" to also including "or just have a folder and get the agent to do it", and I'm finding this pattern pop up more and more for me.
Using ChatGPT or other services to convert the data between various formats indeed saves a lot of time and removes the need to write throwaway code to do that conversion. One should still carefully review the result, because the bigger the size of the data, the more chance of hallucinated results, even on the most recent models; but in general, this is a great improvement to some of my regular data conversion chores.
People need to stop pretending text and voice commands are an evolution on the user interface.
The UI is the innovative thing that simplifies and enables focus - not talking to “an agent”.
Imagine saying or typing what VIN you want instead of a dropdown menu with a list of them - isn’t it way faster and easier to see options and click one in a UI? Rather than have a back and forth conversation?
The take is really - if you zoom out - that we can cram all UX into a chat. Idk about that, I like a UI.
I’m more interested to see what kinds of UI/UX beyond chat we get. If you’ve messed with the AI video SaaS that’s already out there a few UX paradigms are already being established, like the concept of image references - they all do it different ways, but there will be an established way. Like pull to refresh or pinch zoom became established UX for smartphones.
That sounds like a property that's true of good UIs, rather than most UIs, and in particular most SaaS UIs or intra-company UIs. ClickUp's UI is bad enough that I would much rather not use it in preference to chat.
I'd say the inverse - it's only not true when it's a bad UI (like ClickUp's).
I agree though, sometimes text input is a superior UX than using any UI - like Spotlight search or git CLI.
Also agree that AI will broaden that - but the default?
A good thing to keep an eye on is image editing: Will people keep telling a chatbot what changes to make, like a slot machine hoping for the best, or will companies start putting more tools into the workspace (e.g. draw a selection box around an area and "Generative replace...")