AI-generated whatever is everywhere these days. So what?
The real story isn’t happening on Instagram and TikTok. Where AI is having the biggest impact is inside enterprises.
With corporate licenses for platforms like Claude and ChatGPT, highly skilled, sophisticated, and well-resourced internal development teams are now supercharged.
The productivity boosts are massive, with AI doing most of the heavy lifting of code generation. This means internal teams can release custom features at a scale and velocity unimaginable even just a few years ago.
On the surface, AI is delivering the ultimate win for building in-house tech. But scratch beneath the surface…
The Distinction
In the world of software development, there is a huge, canyon-scale distinction between writing code and operating software.
As AI can now handle the vast bulk of the code-writing part, a fascinating (and potentially painful) illusion is being created – that the hard part of creating software has been solved.
Need something. Ask Claude to write it. Done.
However, the reality is that writing code is only a fraction of the process. From our experience, including our experience using AI to supercharge our own code-writing efforts, it’s about 20 percent of the software creation equation.
This is because enterprise software isn’t just written. It’s also operated, so this is where the bulk of the work begins, i.e., after the code is deployed.
In other words, you can’t just write code that works in a sterile testing environment. You also have to make sure it can survive and succeed in a massive, complex, and ever-changing corporate ecosystem.
The Hidden Workload of the Software Lifecycle
So, if 20 percent of the software equation is writing code, what makes up the other 80 percent? The answer: building, testing, and maintaining the operational guardrails across the software lifecycle. Here are (just some) examples of those operational guardrails:
- Deployment pipelines. Deploying code across tens of thousands of endpoints without disrupting operations or the user experience.
- Continuous observability. Monitoring software health, identifying security flaws, catching failures, and spotting performance degradation.
- Compliance. Adapting the software to keep it aligned with changing security and governance policies.
All of a sudden, supercharged internal development teams are not just writing code with the help of AI. They are also in the infrastructure business, so they end up spending most of their time on the plumbing, governance, and guardrails that make the software actually operational in a real-world enterprise environment.
This takes them away from where the real value of AI can be derived – building the actual business logic.
The Strategic Question
In-house teams should be using AI. We use it ourselves at Access IT Automation. It’s a force multiplier like no other.
The question for enterprises is about resource allocation.
Your internal team can innovate at lightning speed using AI tools. That is indisputable.
Should they use that capability to recreate from scratch the operational guardrails required for enterprise software applications?
Or should they focus on unique business solutions? The types of solutions that can only be created by internal development teams using their deep knowledge of the company.
So, building something from scratch that already exists or building unique business solutions and running them on top of proven, enterprise-grade platforms?
Unsexy, resource-consuming work that adds minimal value, or features, apps, and tools that you will never be able to buy off-the-shelf?
Combining Innovation Velocity with Operational Maturity
AI gives internal development teams an incredible engine to innovate at speed. But engines can’t move by themselves. They need a chassis, suspension, brakes, and an electrical system.
Using AI to build unique business solutions that align with enterprise software platforms ensures you have all the necessary components to combine innovation velocity with the real-world operational maturity needed in enterprise environments.
