We've been writing about emerging technology and enterprise trends for the better part of a decade. Some of our predictions landed. Some missed entirely. And some were right about the direction but wrong about the timing. Here's the honest accounting.
What We Got Right
Enterprise Cloud Was Infrastructure, Not Strategy
Back in 2017-2018, we argued that cloud migration was an infrastructure decision, not a strategic one. The strategic value came from what you built on top of cloud infrastructure, not from the migration itself. At the time, "cloud strategy" was the dominant framing.
This landed. The organisations that treated cloud as enabling infrastructure and focused on what it enabled moved faster than those who treated migration as the objective.
Remote Work Would Reshape Enterprise Tech
In early 2020 - before the pandemic - we'd been writing about distributed work and its implications for enterprise tooling. We weren't predicting a pandemic. But we'd observed that the tools and processes designed for co-located teams were inadequate for distributed ones.
COVID proved this dramatically. The prediction was right; the catalyst was unexpected.
NZ Enterprise Needed Local Capability
We've consistently argued that New Zealand enterprises need technology partners who understand the local market - the regulatory environment, the cultural context, the small-market dynamics. This wasn't a controversial position, but it's been validated repeatedly as global platforms and offshore vendors have struggled with NZ-specific requirements.
Data Was the Bottleneck, Not Technology
Going back to 2019, we wrote that enterprise data quality and accessibility would be the constraint on technology adoption, not the technology itself. This has been true for cloud, for analytics, and now for AI. The technology consistently outpaces the data infrastructure.
What We Got Wrong
Blockchain for Enterprise
In 2018, we were cautiously optimistic about blockchain applications in enterprise supply chain and identity. We were wrong. The technology solved problems that enterprises didn't have, or solved them in ways that were more complex than existing alternatives. We should have applied more scepticism earlier.
The lesson: technology that requires ecosystem adoption to create value rarely achieves that adoption in enterprise contexts. Individual enterprise decisions are rational. Collective adoption is a coordination problem.
The Speed of Low-Code Adoption
We overestimated how quickly low-code platforms would penetrate serious enterprise development. The platforms improved dramatically, but the cultural resistance from development teams and the limitations for complex use cases kept adoption narrower than we predicted.
Low-code has found its niche - internal tools, simple workflows, citizen development. But it didn't replace custom development for complex enterprise applications the way some of us expected.
Digital Transformation Timelines
We consistently underestimated how long enterprise digital transformation would take. Projects we thought would take 12 months took 24. Programmes we thought would take 2 years took 4. The technology was ready; the organisational change wasn't.
This is the prediction I should have known better on. Enterprise change is always slower than the technology that drives it. Always.
What We Got Half Right
AI Would Become Enterprise-Relevant
We wrote about AI and machine learning as early as 2018, noting that enterprise applications were emerging. We were right about the direction. We were wrong about the timeline and the mechanism. We expected gradual adoption of traditional ML - predictive analytics, classification, recommendation engines.
What actually happened was ChatGPT: a consumer product that made AI viscerally real for non-technical people and compressed the enterprise adoption conversation from "someday" to "now." The direction was right. The path was completely unexpected.
The Talent Shortage Would Constrain Growth
We've written consistently about NZ's technology talent shortage. The constraint is real. But we underestimated the extent to which remote work would partially address it - NZ companies accessing global talent, and NZ talent working for global companies.
The talent shortage remains acute, but the dynamics are more complex than our earlier framing suggested.
What This Teaches Us
A few meta-observations from reviewing eight years of predictions:
Direction is easier than timing. We were usually right about what would happen. We were frequently wrong about when. Enterprise adoption follows its own clock, and that clock runs slower than technology development.
The mechanism matters. Knowing that AI would become enterprise-relevant is only useful if you also know how. We didn't predict ChatGPT. Nobody did. The mechanism shaped the adoption pattern in ways that being right about the direction didn't capture.
Scepticism ages better than enthusiasm. Our most durable insights were the cautious ones. "This is interesting but enterprise adoption will be slower than you think" has been right more often than "this changes everything."
Compounding beats prediction. The organisations that did well in our ecosystem weren't the ones that predicted correctly. They were the ones that built capability incrementally and adapted when the landscape shifted. Prediction is less valuable than adaptability.
Applying This to AI
So what does our track record suggest about the current AI moment?
The direction is clear. AI will become embedded in enterprise operations. This is as certain as cloud was in 2015.
The timing is uncertain. "Within two years" for widespread enterprise AI is probably optimistic. "Within five years" is probably conservative. Somewhere in between.
The mechanism will surprise us. The next major shift in enterprise AI won't come from where we expect. It never does.
Build capability, not predictions. The right response isn't to predict accurately. It's to build the data, the skills, the governance, and the organisational readiness that let you move when the opportunity arrives.
That's what we're doing. And we'll keep being honest about what we see - including when we're wrong.