>> Tools you pay for that feel overpriced
I get your thinking here - free market spots a gap and competes on price. But personally I think this is a "poor" business strategy.
Firstly, competing on price is a race to the bottom. There's always someone who's prepared to offer it for less. And that always means the offering ultimately gets worse over time.
Secondly if you go thus route you attract customers who are "price sensitive". They'll leave you as quickly as they joined for someone who offers a lower price. In other words you are self-selecting the worst possible kind of customer to have.
Thirdly you send a signal regarding quality. A $5 product is obviously worse than a $100 product, because if the $5 -could- charge more they -would-.
(Hint: for VC funded companies the customer I the VC NOT the person paying a subscription. So be careful comparing your pricing to VC company pricing if you are not VC funded.)
Yes, look for pain. If you make something good charge more, not less, than incumbents. Build a customer base of people who are quality sensitive not price sensitive.
30 years ago I had a product on the market at $199. It sold well, validating the market space. A not-quite-as-good competitor appeared for $99. He made some sales. I was tempted to reduce my price. An old head told me otherwise. Sales wise we'd say things like "you get what you pay for". We outsold the new-guy. Which meant we had lots more revenue for support, docs, polish and so on. Today we charge $399. The competitor folded after a couple years.
We aimed high, charged a lot, and focused on delivering quality. We attracted loyal customers who appreciated quality more than a few $ saving. 30 years, and 50 products later, we continue to dominate our space.
I agree that driving down pricing gets you customers you don't want, but LLM-written software changes the game. If it cost $100,000 to build a thing before, and now it costs $10,000 to build it with AI, your upstart competitor can charge $50 for something good enough to be competitive to your $199.
Maybe. But maybe not.
Firstly, writing the code is perhaps only 10% of the product cost (over its lifetime.) Support, marketing, adjusting, debugging, handling edge cases and so on are where most of the costs lie.
In my case brighter folks than me wrote very good code in the same space. But taking working ode and turning it into a generic product is a lot of work.
Think of it like building a landing strip for your plane. If you're the only one landing there, then the specs are simple, and tight. You need little more than a grader and a building big enough for your plane. But if you're building an airport (think, say, JFK) you have to cater for everything from a Cessna to a Dreamlifter.
A LLM can certainly speed things along. But I wouldn't want to build based on LLM foundations. Long term support and maintainability would scare me.