Lack of automation AROUND AI kills generative AI gains

The deeper and more serious I am getting into personal AI, the more obvious it has become to me that handling the more stuff I am generating with AI is eaten up by its administration: faster reporting gets more reports. Greater access to information results in more information to tag, archive, describe, organize, handle... so that it is not lost after a few months. More files. Bloated Databases. Many different AI and non-AI systems to feed.

And all this meta-level stuff cannot be optimized by AI because it's all about NEW creative stuff, projects, organizations, etc. AI has no role model it could have learned from, and it's not the kind of CONSISTENT and STRUCTURED data current AI can learn from at all.

So, finally, I ended up with more that has taken a lot of additional time to keep it actionable and relevant for the long run. Or to put it bluntly, this increased productivity on single tasks did not earn me a dime. Not by money, but even worse - not by time, because all these gains were immediately eaten up on the meta-level where the "more" cannot be absorbed the old way. When just parts of the overall-supply-chain is on warp speed, your business is not, as overall delivery speed is still defined by the slowest link.

So, I have put my focus on the "classy" automation of everything AROUND AI output, starting with CopyQ Clipboard -Manager, which has turned whatever I copy from AI output into a real-time dispatching and journaling machine that automates the "post-production" handling of AI generated output - and here is the banana- I do this kind of automation the old way. Not AI, but classy automation stuff. And only THIS has turned my AI-supported production stuff into a real powerhouse.

If we just create more AI output that doesn't AUTOMATICALLY show up in and work towards its final pay-channel, we are not there. We are just bloating parts of the machine, but not production as such.

However, making such full automation a reality, is according to me, still old-school automation business that is not learned over the weekend.

And so, I am wondering whether AI will - on the long run - only benefit those who can automate their business end-to-end, which necessarily will require heavy automation skills.

Maybe AI gains are not really cash cows for those who have never automated anything. To cash in on AI still required at least an engineering mindset. If you do not know what this means, keep AI for fun, but do not expect to get rich doing so.

This is especially true when we go towards agentization. Creating value on this is more about dealing with classy automation complexity than just prompting. And vibe coding will not do the trick. I would have been the first using vibe coding. And I used it a lot. And my lesson on this is that vibe coding is a great assistant for great developers but will never ever come with readymade implementations that will earn you a life. And the worst is, that many organizations are laying off automation experts, project leaders, change managers, etc. with incredible know-how about classy automation without realizing that exactly these automation skills will be paramount to using and earning from AI's full potential.

2 comments