Should AI products let users control how much freedom the AI has?
While working on an AI product recently, I ran into a product-design question that turned out to be much more important than I expected: how much freedom should the AI have when the user gives it constraints?
I initially assumed that better AI meant giving the model more freedom to find the best answer. But in practical workflows, that isn t always what users want, and sometimes the constraint itself is the point.
When do you decide to move ahead from your product?
I have made two different products (none worked). When starting with the first product, I thought of marketing it through social media and paid ads. But, a few weeks later, I came to realize that the market I am trying to make my customers are not really a good fit for the product and that the market is not having this problem yet. So, I moved from it after a few weeks.
Then, with the second product, it was a completely opposite situation. I created a product which had extremely competitive market and several other devs had created the same product as mine or even better once (regading automation). A few weeks later I moved on from that.
One lesson learned, always do market and competition research. When do you decide to leave the product and conclude it is not worth it now?
How do you actually verify a "track record" before trusting someone with your money?
Curious how this community thinks about trust/verification problems in general not just in my space (forex), but anywhere someone claims a track record (traders, freelancers, agencies, etc.).
I've been building in the forex space, where the standard "proof" is literally a screenshot of a trading account easy to fake, easy to cherry-pick, no way to independently verify. It made me think about how many other industries still run on unverifiable claims by default, and how few platforms actually build verification into the product rather than bolting it on as an afterthought (or not doing it at all).
Question for fellow developers: can anyone code with AI?
Honestly, no. Not fully, especially if you've barely gone past "Hello World." Even with AI, one tiny bug can eat your whole afternoon, and a feature that sounds simple can take a week when you don't really know what you're looking at. You also need real exposure to how these coding agents actually work, or you'll burn through money fixing the same mistake five different ways before you figure out why.
AI closes the gap. It doesn't remove it.
What's your experience been like? Did it get easier once something clicked, or is it still a grind?
Is mainstream self-hosting actually approaching, or will it stay a niche?
I've noticed a growing undercurrent of people wanting real ownership over their digital data, their photos, notes, bookmarks and media, rather than renting access to it from a platform. Sony ending physical disc support on their next console reinforced this for me, since it's another case where you own less of what you paid for and control less of it too.
At the same time, people generate more personal data every year, and NAS and self-hosting tools have never been more capable than they are right now.
So I'm curious what this community thinks. Is there a real path for a genuinely consumer friendly self-hosting product, something as easy to own as a router, to actually go mainstream? Or does the setup and maintenance overhead mean this stays a niche for people who enjoy tinkering, no matter how good the software gets?
Is the sign-up requirement what actually kills shared lists?
Disclosure first: I build a shared-list app, so I am not neutral. But I keep running into something I would like other people's read on. Every shared-list tool I know asks everyone in the group to install it and create an account before the list works. In my experience that is where sharing dies not because the apps are bad, but because the person you are sharing with has to do three things before they can tick off "milk". Most people give up and paste the list into a chat message instead.
So my question to people who have built collaborative tools: have you found the same, and what did you do about it? Did you make accounts optional, defer them until someone needed sync across devices, or did you decide the account was worth the drop-off because of what it enables later?
Does anyone else try a tool, like it, and then completely forget about it?
This happens to me all the time. I find a new app, use it for a few days, like it, and then just stop opening it. Nothing was wrong with it, it just quietly dropped out of my routine.
Even great tools fall through the cracks because the habit never formed.
Does this happen to you? How do you get a new tool to actually stick in your workflow?
Does anyone else have a "maybe later" list they never actually touch?
I have a habit of seeing something cool, telling myself i will check it out later, and then completely forgetting about it. It is not even an actual list, just a vague memory that i saw a neat tool once, with no idea what it was called.
The annoying part is knowing there are probably a few useful tools floating around in my head that could help me right now if i had just written them down.
How do you handle this? Do you bookmark things, drop them into a quick note, or just let them go and trust you will run into them again if they are actually worth finding?
My Problem With Productivity Apps.
ok so As a college student and an ex-aspirant who s spent way too many hours preparing for exams half the time in exam prep went into managing my schedule, fighting procrastination, and staying away from my phone while trying to focus.
And I already have reminders, screen-time limits, to-do lists and schedules.
But they just don't really know me ,they were built around fixed schedules and generic plans, while my actual day was never that predictable.
So :-
We spent 20 years optimizing marketing for human attention. What happens when the audience is an AI?
More of what gets "read" online now isn't read by a person at all. It's summarized by an AI Overview, cited by a chatbot, or scraped by an agent doing research on someone else's behalf. The content still exists, but the human eyeballs it was written for are increasingly optional.
That's a weirder shift than it sounds. Marketing is built almost entirely on models of human psychology: attention spans, emotional triggers, scarcity, social proof. None of that obviously applies to a language model deciding whether to cite your page. We might be optimizing skillsets that are quietly becoming irrelevant for a growing share of the funnel, without really noticing.
I don't think anyone knows what "persuading an AI" even means yet, or if that's even the right frame. Maybe it's not persuasion at all, just legibility: structure, clarity, verifiability, instead of appealing to a reader's ego or fear of missing out.
Curious if anyone here has actually felt this shift concretely rather than theoretically. Has anything about how you write or structure content changed because you're now writing for a filter, not a person?