The bottleneck isn't ideas or the AI tools — it's judgment. Knowing which problem is worth solving, for whom, and whether what you've built actually solves it. That judgment can't be generated; it has to be earned.

Vibe coding lowered the barrier to shipping code, but it didn't lower the barrier to shipping value. The result is a lot of technically impressive things that nobody needed. The first comment here (lots of tiny personal tools that never get released) is actually the realistic picture: most of what gets built with AI tools is either too narrow to scale or not quite right for anyone but the builder.

The missing discipline is customer value validation — which is why it's the first principle of the Agile Vibe Coding Manifesto (https://agilevibecoding.org): "Acceleration in development must translate into validated customer value. Speed without value is waste." That sounds obvious but it gets skipped constantly when AI makes spinning up new projects so effortless.

The real unlock will be when people pair AI-generation speed with the harder work of understanding user problems before building.

A VC backed tech is focused on ideas that can go to a billion. A billion dollar idea was always worth pursuing. There is no reason that AI would result in significantly more of these ideas.

AI really enables all sorts of tiny stuff that would never have been worth building and for starters, is not really worth maretking. I have a dozen tiny pieces of software that are used by myself or maybe up to 10 people that that I don't push around or market or release in any meaningful way.

I built a tool to make HTML driven PowerPoints. Claude is great at beautiful HTML, so why not just do the slides as HTML convert them to images? An absurd project before given that HTML used to be work to write, but it is no longer work, making it a great slide design language.

I sell flight deals as a side hustle. Built an automated pipeline so I am just dropping in the link.

I built a tool for a PM friend to let Claude Code automatically sync all his meetings and let him query his transcripts for action items and give him advice on what to do about certain things. Nobody else uses that.

I built another tool for a PM friend that is basically AI for product managers who want something more structured than Claude Code and don't want to handroll their stuff. He and a few others love it, but there's been no attempt to sell it or anything.

I built an accreditation management system for a university here in Canada with a friend. We could try and sell it, but we have no idea how to do that really, haha, so it ended up being a one off to replace a more expensive SaaS subscription.

I have a custom event discovery calendar where I have a database and AI try to fill in empty spaces where I am interested in doing stuff. Again, no users other than me, but great for me.

> What do you think the real bottleneck is now? Ideas, distribution, taste, persistence; or something else?

Probably someone to take it to market. I am sure you could find customers for some of the things I have or at least users, but idk how to do any of that.

The real bottleneck to me seems like you mostly gave superpowers to the engineers, but none to the salespeople. Nor did you give the engineers problem discovery and go to market superpowers, so lots is getting built but nobody is really pushing it out there.

With AI, I've tackled lots of personal projects I had on the back burner for lack of time: recreating some old games, creating some marketing and shopping sites for the wife and also professionally: using it to fulfill business requirements. So in my little corner of the world it is being used and a lot mostly nonstop from Aug 2025 to present. Re your question: I think it has to do with all the rest of what's needed for a production system, the documentation, CI/CD, production deployments: all those pesky still-needed-human-things that are somewhat tedious to set and get correct.

AI can handle all the docs, CI/CD, and deploys too if you give it CLI access and terraform.

The bottleneck isn't usually coding speed. Most of the AI-generated stuff that doesn't turn into useful software fails before the first line is written — nobody actually understood the problem they were solving.

The projects that do ship and matter tend to come from people who had a specific, painful problem and built exactly for that. The vaccine appointment site you mentioned is the perfect example — someone was solving their own family's real problem, not exploring what AI could generate.

What kind of software were you hoping to see more of? Curious if there's a specific gap you're noticing.

Thanks for commenting.

The projects that I always admire here and out there is something like this:

1. someone find real problem (like very tedious to use gov site)

2. build a solution in a weekend

3. solve the problem for themselves and the rest of us.

Bonus: The world is better than it was.

Another very nice project I admire and is recent one is this one [1] and this one [2] as you can see it's to the point and very clear it solves a specific problem and it's useful for a lot of people. Does not necessary to millions but it's useful to enough people nevertheless.

My question is then if AI is that mighty why I don't wake up to 100s new and very useful and exciting side projects and nice tools like this every day?

Other question: why we have an investigative journalism that's is its very business model based (partially) on the incompetence of a "byzantine" fed system? [3]

Again if AI and its Agents are that mighty; why I can't task them to guide me through a byzantine system. Didn't you (not literally you but those who say AGI is almost here) told me that AI is better than the best lawyer and the best investigative journalist and most talented programmer?

The point here is if AI is that mighty in developing software why we don't have solutions for incompetent/ insufficient tools/platforms gov or otherwise.

For me here's a few examples of tools I want to see:

1. Tools that I can connect directly to my PayPal/Wise/RedeotPay/Binance accounts and give me jargon-free actionable insights. No setup needed. No CSVs download/uploaded needed.

2. Tools that I can upload to them my FB/LinkedIn/Twitter/Gmail/WordPress archive and give me useful psychological/marketing related/productivity/ideations/patterns's unearthing insights.

3. Same as 2 but for my browser data and bookmarks and history.

To answer your question directly:

> What kind of software were you hoping to see more of? Curious if there's a specific gap you're noticing.

the gap I see is Useful analytical tools that when I give it already existing data (may it be a jpgs of My handwritten journal or my finances CSVs or My search history...) it give me useful insights that make me able to do more of what works and less of that doesn't.

More indie projects I admire and use:

1. yakread.com

2. blogtrottr.com

3. f5bot.com

4. www.newsminimalist.com

5. Kagi News

6. Bluesky Directory

----

[1] site about deals by Jenny Ouyang

https://buildtolaunch.substack.com/p/app-worth-building-ai-v...?

[2] Substack Reader Chrome Extension by Karen Spinner

https://open.substack.com/pub/wonderingaboutai/p/i-built-a-c...

[3]

https://seamushughes.substack.com/about

Read:

> A bit about Court Watch. We have a weird knack for finding the unfindable in the byzantine federal court records system.

> Additionally, for a limited time, if you become a founding member, you get an online zoom training by us on how to traverse the federal court system.