Moral Support at the End of the Road

Hello. Happy Halloween, if that’s your thing. I’m not really someone who likes to be doing things because the calendar says so, like eating pancakes or getting people presents. It’s okay to do those things any day you fancy! Of course, you’re always welcome to hop into my calendar.

There’s trouble on the horizon and I think we can all see it looming large. The simple fact is that none of these big AI products are approaching profitable and building larger and larger towers atop weak foundations isn’t going to lead to good results. Soon OpenAI and their competitors will run out of other people’s money and all these apps which are essentially GPT and Claude frontends (things like Cursor, all these copilot features) will be so expensive as to be unviable.

My feeling is that the trigger will be a spike in electricity prices. The expansion of renewable generation has been constrained and there isn’t an end to global instability in sight, which always has a large effect on energy prices. If the cost of electricity doubles then I don’t want to be sat there funding huge datacentres.

OpenAI and Anthropic based tools have embedded themselves in the customer operations world like knotweed. Good old fashioned deterministic engineering has been shunned in favour of language models which are effectively being subsidised. Have you noticed Netflix and Spotify and Uber prices shooting up way ahead of inflation over the last few years? These businesses did the “do it at a loss to get people hooked, then make the money back later” thing. OpenAI is going to have to pay its debts at some point.

I simply cannot recommend products to my clients which aren’t backed by vendors with a viable business model. In the past I have been known to work on migrating to a tool which subsequently was mothballed and the client unexpectedly had to rework processes and spend more money solving the same problems again. If I don’t think a tool will exist in a year or two or if the pricing is subject to too much volatility, this is something I will steer clients away from.

Yes, there is trouble coming. You know how in the US it’s only a few companies who are really growing? Take away Meta (Facebook), Alphabet (Google), AWS, Nvidia, Apple, Microsoft, and it’s clearly a massive recession. What is the situation if you’re a CEO of a public company, not one of these titans but still a big business?

The coming trouble is not directly a huge panic for the people who make the decisions. CEOs of regular publicly listed corporations are effectively measured against the growth of the index as a whole. If you’re running a FTSE 100 company then you can’t be growing slower than the FTSE 100 average or the board will fire you on behalf of the shareholders. So you have to do the things that make your company look like it’s doing the things that the successful businesses are doing so that the share price is steady. I appreciate that sentence is ugly. No AI in my writing.

No AI here, hand drawn pixel art and SVG logo, just like mother used to make

That’s why you get crazy announcements like the brands we have seen saying they have replaced 1000 humans with a new ai bot. it’s just nonsense but they have to do these performances or people realise the business is overvalued, the share price drops, the board of directors fire the CEO and then someone else is installed who will do these things. AI is the ketchup you can pour on to hide the fact that you burned the dinner. If you have a nice meal, you’ll only want a little dollop of it. It’s the same with your business processes.

I keep saying the same thing to anyone who asks (and some people who don’t) because many were saying AI is a gold rush. Who made money in the gold rushes? The people who sold the shovels. Everyone else was basically just scrabbling around in the mud. Nvidia will be the winners. Apple too, because they haven’t been drawn into competing with ChatGPT but they do make the devices everyone working on these things wants to use.

What is the conclusion?

We are going to stay the course. We believe that good processes and proper engineering will always be in vogue and we’ll probably be busier than ever when these AI wrapper products start failing and businesses realise they always needed real solutions.

We’re going to renew our focus on bringing our own products to market which exemplify these values. You need to know what a given input is going to produce as output. You need to be designing flows, not cargo culting prompt engineering. It’s the end of 2025, we all know what slop looks like. There was five minutes of fun while we all saw what we’d look like if we were characters in Ponyo or we were Mattel action figures. Now let’s get some work done.

What does this mean for Moral Support?

Everyone has stopped reading newsletters, I think because everyone has stopped writing newsletters. There are a lot of them bouncing around, but they are all AI written wordy gibberish. I think the people have spoken. If the internet is full of content humans didn’t want to write, then it has become a place humans aren’t looking for things to read. I know I get angry when I get an AI-written email (usually a hundred or so a day, so it’s good I’m fueled by that) and so the newsletter format just seems to have died.

My perspective is that the best way to be a real human on a dead internet is to do things which are unfiltered, raw, live, imperfect. That means this newsletter is probably over but I’m going to play with other formats. We should show off the things we’re doing, so we’re going to be building in a more public way. Sometimes we’ll get things wrong. We’ll definitely be human, and we’ll definitely be getting stuff done. Watch this space! Coming in the next few days is a great chat we did with a top vendor who we do recommend to clients.

Expect to see more of me through this