Reading between the bank holiday headlines this weekend (May 2026), we’d be forgiven for believing that the machines have already taken over. That the tech bros are already carving humanity up between them and that large scale AI-related mass unemployment has already happened in all sectors all over the place. In other words that we’re all already fecked basically!!
But wait. For the few of us who’ve ever been involved in looking into what it really takes to get Agentic AI processes and projects off the pipe-dream pages and into reality, a slightly different picture emerges.
First of all, have you ever tried getting definitively sound, unquestionably expert advisory on AI data, automated workflows, cybersecurity or agentic design? Not simple. You get so far, you think you’re away with it, then you discover some new detail which contradicts one of your core AI beliefs which you had adopted as gospel, so now you’re straight back to the drawing board.
Second, once you do decide what your AI jump-off point will be, finding a proper implementation partner with the right expertise and experience and at the right price point, is also quite the challenge.
That’s all before you get to the much-commented-on failure rates in AI projects. Before you get to the technical challenges of ensuring your data is secure and not stored or processed outside of the EU. Before you get to the ongoing cost of buying AI tokens to maintain whatever AI processes you do develop and implement. Before you have to contend with staff who might not yet fully understand what the hell this “agent team” actually does or why it’s needed, or why it’s taking up so much time to get set up in the first place. Before the now recognised AI brain fry takes hold in the poor unfortunate charged with maintaining managing the AI agent team.
So you’ll forgive me, with all these real-world pitfalls, if I’m more than a little sceptical about just to what extent AI has really become a factor yet in our everyday economy. Yes, there is a movement well underway. Yes some of the AI tools such as Claude Opus are absolutely mind-blowingly good. Yes, there is an employment numbers correction in the technology sector. But there are even questions about whether that really is ALL as a result of AI and not just conveniently proximate to AI so that that becomes the much-needed excuse for rationalisation.
Regardless of where in the technology cycle we really are just now, it remains the case, that there is simply no substitute for that human trust, that human connection and those real business relationships we all depend on to advance our AI projects on to the next level.
Of course machines can help us with the prior research. But that only gets us so far. Eventually we need stable, trusted peers around us to help us fill in the gaps in our knowledge which, let’s face it, because it’s all so new and developing so quickly, most of us are naturally still a bit at sea with the whole Agentic thing.
So let’s just start there 🤔 Whether that be recognising that new AI skills and technology talent needs to be hired in, or whether it’s new network connections to be sought out and developed. There is absolutely nothing artificial about the human factor needed to get these things just right, it’s still all about people. Way more than just the “human in the loop” as the requirement suggests.