We’ve started talking to computers.
And it’s already changing what software is.There's a scene in Star Trek IV where Scotty picks up a mouse and says 'Hello, computer.' It was a joke, of course. At the time, you adapted to machines, learning their commands, their syntax, their structural logic, because they couldn't learn yours.
Now it's the other way around, and the reversal is more consequential than it first appears.
We can now (or will very shortly) become habituated to asking questions that cut across systems, workflows, and job functions and then expect something usable back - no training, no manual, no ‘learning to work the machine’. That shift, from operating software to talking to it, is already changing the industry, quietly but consistently, in the way that the most significant changes tend to happen.
Parts of software are beginning to look exposed, not because they're fundamentally broken, but because the use case or user expectation has changed. It's no longer 'what tool do I need?' but 'can I just ask for this instead?' - and in more and more cases, the answer is yes. You experience it before you really notice it: small tools quietly dropped, internal teams building what they need in days rather than procuring it over months, budget conversations opening around products and long-running partnerships that were never up for question before.
Underneath this change sits a slightly uncomfortable truth, which is that some of this software wasn't valuable for any other reason than it was the only practical option at the time, and AI has changed that. When the interface disappears, and the workflow compresses into asking a simple question, you find out quickly what was doing real work and what was simply sitting in the middle of a process, necessary only in the absence of anything better.
The deeper systems, meanwhile, aren't going anywhere - if anything they're becoming more important, because AI needs something solid to work with: data that's consistent, comprehensive, and grounded in how the business actually operates rather than how someone once modelled it (economists call this difference “expressed vs revealed’- the first is what you say or think happens, the second is what actually happens). Without that grounding in reality, AI doesn't help or create real clarity; it just produces faster answers to the wrong question. That is in many ways a different problem (and likely a worse one).
In maritime, that distinction matters more than in most industries. Commercial, technical, financial, compliance data - all moving simultaneously, across systems, teams, and time zones - which is exactly why the idea of a single, connected operational picture has always been a need. When someone asks for a number, whether that's cost, position, or status, it needs to be right; not close enough, not assembled from several different places and hoped to cohere, but right. AI doesn't remove that problem- it elevates it in criticality, because now that overall view is within reach, but it must be right.
Where that leaves SaaS is probably in a more honest place, less concerned with features and dashboards - the things easiest to see and easiest to sell - and more with the parts that are harder to build and easier to overlook: how the data sits underneath, how cleanly systems connect, whether the output can actually be trusted when a decision depends on it.
The question is changing too. 'What can it do?' is giving way to something more direct: 'Can we trust it?' - and that is a different kind of question, requiring a different kind of answer, from software built differently than most of what currently exists.
The shift isn't announcing itself in the way that, say, the arrival of the Internet did. What used to feel like science fiction - talking to a machine, asking it something complex, receiving a meaningful answer back - didn't arrive with a moment of revelation; it became (almost without anyone deciding so) just how things must work.
In other words, the value in software is moving away from the surface experience and deeper into the system, into the data and the decisions, and the place where the truth actually lives. The companies that understand this sea change early will have a significant advantage over those still selling the interface.
John Wills
COO, Shipnet