dug 
So have started a podcast. Sort of.
And as a result, have been getting recommendations from all points on AI-enhanced services and automation.
It turns out that making podcasts is actually reasonably complex from a content design point of view. You need to make “show notes” (or course you do) and “clips” (long-form for Youtube, short-form for other stuff) and you have to write all sorts of snippets and widgets and doodahs to promote your show. Then, you need to interact with Apple, always a painful experience, and discover that their podcast hosting tool doesn’t let you change or edit anything once you’ve published (not good for a suck-it-and-see beginner) nor, for that matter will it let you delete your crappy test broadcast…
Which means that guess what, the ChatGPT-powered start-up universe is on fire with services that automate some or most of your production and marketing workflows (see https://www.thepodcasthost.com/planning/ai-podcasting-tools/). Some of the tools are VC-backed (pretty sure I’ve seen a Y Combinator logo on my travels) which I guess isn’t surprising given all the AI hype, but from a total addressable market perspective, who knew there were were that many potential customers for this in the creator community?
But what I wanted to whinge about wasn’t the podcast market or AI but the design choices around billing and paywalls in these systems.
Specifically, how are the founders of these services defining product-market fit? What are the pain relievers meeting the pains? What, exactly, is the customer renting? What is valuable to the customer and are they getting it?
I had an email exchange with the (charming and friendly) founder of one of these services where we discussed their struggles to get the paywall right. I had signed up for their service as it it looked great, and almost immediately upgraded to a paid account as I figured this was the service that was going to fast-track me to where I needed to be.
As I struggled to get my episode ready to publish, the paywall came down (it said I’d used up all my credits and needed to upgrade to continue).
A bit later in the day I was trying another service and a similar thing happened (“Purchase minutes for $0.15/min”).
And of course Dalle-2 used to sell units as well.
So my feedback to the podcast service was really why was I paying them? Was I sub-letting units of AWS compute time from them? Or, was I renting the ability to be a podcaster?
(Because frankly, there are much better way to pay for cloud compute cycles)
Anyway, feeling quite frustrated as I’m finding that in this AI-startup world there isn’t currently a correlation between buying from or subscribing to a service and receiving unique value in return. That is, I’m paying money, and not getting the thing I needed to get my job done.
Hopefully, as the market matures a bit this will begin to make more sense…