I make a habit of writing down predictions every few years. Not because I'm particularly good at predicting the future, but because looking back at where I was right and wrong tells me something about my blind spots. Here's what I think the next five years of enterprise technology look like from October 2022.
What You Need to Know
- Integration platforms will become the centre of enterprise architecture, not the periphery
- Cloud computing will mature from "move everything to the cloud" to "use the cloud strategically"
- Automation will expand beyond RPA into more intelligent, context-aware workflows
- AI/ML is progressing faster than expected, but the enterprise applications remain unclear. Something significant is building, but predicting exactly what is premature
Integration Becomes the Centre
For most of the last decade, integration was treated as an afterthought. You built your systems and then figured out how to connect them. Over the next five years, I think that inverts. Integration architecture becomes the starting point, not the final step.
The reason is scale. NZ enterprises that had five systems in 2018 have twelve in 2022. By 2027, the average mid-market company will have twenty or more software tools in their stack. The integration complexity ceiling we wrote about earlier this year is going to hit more organisations, harder.
1,061
average number of SaaS applications used by large enterprises, up from 599 in 2018
Source: Productiv SaaS Trends Report, 2022
Integration-platform-as-a-service (iPaaS) tools will grow. Middleware will get better. But the bigger shift will be architectural: organisations will start choosing tools based on how well they integrate, not just what they do. The best tool that doesn't connect to anything else is the worst tool.
Cloud Strategy Matures
The first wave of cloud adoption was about migration: move your stuff off-premises. The next wave is about strategy: figure out what belongs in the cloud, what doesn't, and how to manage costs.
Cloud costs have surprised a lot of organisations. The promise was "pay for what you use." The reality is that usage grows silently, egress charges add up, and the effort to optimise cloud spending is a full-time job. Cloud cost management will become a standard enterprise function, not an occasional audit.
I also expect to see more organisations adopt multi-cloud or hybrid approaches. Not because multi-cloud is elegant (it isn't), but because the risk of single-vendor dependency is becoming clearer. The cloud providers know this, which is why they all invest heavily in services that create switching costs.
Automation Gets Smarter
RPA (Robotic Process Automation) had its moment. It works for well-defined, repetitive tasks. But the things enterprises most want to automate aren't simple rules-based processes. They're workflows that require judgement, context, and the ability to handle exceptions.
Over the next five years, automation will get better at handling these cases. Not through general artificial intelligence, but through better workflow engines that combine rules, simple ML models, and human decision points. The automation won't replace judgement, it'll handle the 80% of cases that are straightforward and route the 20% that need human attention.
The gap between what automation can do and what enterprises want it to do is closing, but it's closing from the automation side, not because enterprise problems are getting simpler.
Isaac Rolfe
Managing Director
The AI Question
Here's where I'm least certain, and most interested.
The pace of AI progress in the last two years has been faster than I expected. GPT-3 was impressive in 2020. The image generation models that emerged this year are impressive in a different way. The research papers coming out of Google, DeepMind, and the major labs suggest that the next generation of AI capabilities will be another step change.
But what does that mean for enterprise technology? Honestly, I don't know. The demos are compelling. The applications that actually work in a business context are limited. There's a gap between "AI can write a reasonable email" and "AI can reliably handle a customer service interaction," and that gap involves trust, reliability, auditability, and accountability, all things that enterprise requires and that current AI doesn't provide.
My sense is that something significant is building in the AI/ML space. The investments are enormous. The talent migration toward AI research is real. The capabilities are improving at a non-linear rate. But I can't draw a clear line from where we are today to a specific enterprise impact. It's a feeling more than an analysis, which makes me uncomfortable writing it down. But here it is.
If I had to guess, I'd say the enterprise AI moment is two to four years away. Not general AI, but specific AI applications that reliably do things that currently require human cognition. Document processing, decision support, pattern recognition across large data sets. Tools that augment human work rather than replace it.
What I'm Not Predicting
The metaverse as an enterprise platform. I know Meta is spending billions. I don't see the enterprise use case beyond niche training and simulation scenarios.
Cryptocurrency and blockchain for enterprise. Still a solution looking for a problem in our market. I've been wrong before, but I don't see this changing in five years.
Low-code replacing development. Low-code tools are useful for specific applications. They don't replace bespoke development for complex enterprise systems. They'll complement, not substitute.
Checking Back
I'll revisit these predictions at the end of 2024. Based on my track record, I'll be roughly right about the integration and cloud trends, partially right about automation, and either dramatically wrong or unexpectedly right about AI. The things I'm most uncertain about tend to be the things that change the most.