Each time such a technology is introduced, the easiest way to adopt it is to optimize steps within existing processes. Such optimization is easier as it requires less change - a team can take a process they’re already familiar with and make it better with the new tech. Furthermore, such change is local - meaning other teams don’t need to be involved, so the change is easier and faster to perform. Furthermore, it’s done within familiar boundaries - no big changes to org structures, ways of evaluating performance, ROI calculations, fewer approvals required… easy.

The last disruptive software tech shift before was the cloud. When introduced, companies were reluctant to adopt it. Over time, the industry has grown to appreciate that the cloud is immensely valuable, and that adopting it is necessary for anyone seeking to remain competitive. Eventually, we understood that true adoption of the cloud requires a change to the way we work, and companies routed huge budgets towards “Digital Transformation” projects that let them not just use this technology, but adapt themselves to it - and truly level up.

The same is true for AI. Some companies are in denial, avoiding the use of it entirely, perhaps due to security, reliability or other concerns. Others are actively seeking its adoption, trying to find ways AI can automate and streamline their current practices.

Let’s review the three stages companies go through in embracing a new technology - Denial, Adoption and Transformation - and compare what happened with cloud to what is and will happen with AI.

Denial & Deferral

At first, companies were either afraid of using the cloud, or dismissive of its benefits.

The general wisdom amidst big companies was that the cloud was for smaller companies, who don’t have real scale or compliance needs, and that it wasn’t “enterprise ready”. Beyond the maturity concerns, there was lots of pushback around security, claiming responsible companies would never send their data outside their systems - how irresponsible! Lastly, there was lots of doubt about the future potential of cloud, or whether it was really novel. Larry Ellison’s tirade claiming cloud is just a fashion trend is especially memorable.

This approach kept organizations from making the effort required to adopt the cloud early. Such early adoption was indeed harder, as this was new technology with many warts. However, the delay in adoption allowed early adopters of the cloud (mostly startups) to leverage its fundamental benefits and threaten the slow moving giants.

With AI, we’re seeing a similar narrative play out - though it’s playing out at faster speeds.

You’d find fewer people claiming AI isn’t significant, but it’s easy to find claims it is over-hyped. It’s even easier to find statements about the immaturity of the technology, and how it’s not fit-for-use in the enterprise. To my ears, the concerns around data security, reliability and lack of surrounding features sound awfully similar to the early cloud days…

To be clear, many of these concerns are entirely legitimate - this is new technology and it’s clearly immature in many ways. However, those who don’t make the effort to overcome these shortcomings today are likely going to suffer the same consequences as the laggard cloud adopters did before them.

Adoption & Automation

When companies chose to adopt the cloud, their initial use was naturally within their existing workflows. This was no coincidence - AWS intentionally chose S3 and EC2 as their first services, as they were the easiest to adopt. Companies started offloading storage to the cloud, and eventually started “Lifting and Shifting” workloads to the cloud - run the same VMs in a cloud hosted environment.

This approach makes sense. Replacing on-prem VMs and storage with their cloud peers removed the need to manage capacity, allowed for faster response to changes in demand, and reduced data center operator costs. At the same time, it required relatively little change to the development process. For example, there’s no need to retrain staff, or rewrite existing apps… an easy win.

AI adoption is following the same pattern.

Coding assistants save developers from typing code they were already about to type. ChatGPT makes it much easier to use existing knowledge than finding and adapting a similar StackOverflow snippet; AI test generators write the unit tests that developers would - or should - have written anyway; while AI documentation generators spare them the necessary, but often forgotten, updating of the docs.

These tools are excellent productivity boosters. No change to workflows, no retraining of staff, no rewriting of apps. The tools don’t work reliably yet, but when they do - they are provide great upside with little cost.

Button "Here more about AI native development"

True Transformation

But of course, the cloud journey also showed us the story doesn’t end there. To reap the true benefits of the cloud, you had to embrace cloud native.

Cloud native development brought a different approach to building software, redefined from first principles with cloud capabilities built in. It introduced previously unseen practices such as continuous deployment, realtime elastic scaling and immutable infrastructure. It thrived when rewriting apps to use container-based microservices that can be deployed and scaled independently. It introduced the need for new tools, such as cloud-focused observability and security posture management, and a new flavor of infrastructure as code.

These changes bore fruit. Cloud native, DevOps-powered orgs have repeatedly proven to perform much better. They gained a faster time to market, higher resilience, lower operational costs, higher employee retention, and more. However, embracing these new workflows required A LOT of change. Startups could effortlessly embrace a cloud native approach from day one, but existing companies had to invest heavily in “Digital Transformation” to make the leap.

The question is: What does the AI equivalent of cloud native development - AI Native Development - look like?

What would emerge if we redefine our approach to building software from first principles with AI capabilities built in?

And lastly, what advantages would such AI Native Software Development bring along?

Our conviction is that the opportunity to improve software creation is at least as big - and likely much bigger - than the one the cloud brought along. We intend to work with the brilliant dev community to define what AI Native Software Development is, and to build a platform that enables building software under this powerful new paradigm.

To get involved, or just to hear how we’re progressing on our journey, sign up with the button earlier up the page, or the form over on the right hand side and we'll share more news with you when we have it. Alternatively, follow us on social and our podcast via the links in the sidebar (below on mobile).