“Whether or not you want to actively participate in the AI revolution, it will happen around you. The way API-first architectures did. The way cloud adoption did.”
— Leda Glyptis

‍

$1.52 trillion is a big number. So is $520 billion. Or anything that costs $3.88 million per hour.

‍

These are some of the headline-sparking numbers from a 2022 CQSR report, quantifying the ball and chain of technical debt ($1.52 trillion), of which $520 billion is pinned on legacy systems. These Boomer and GenX systems are aging worse (arguably) than their human counterparts: the cost of an IT outage is estimated at $3.88 million per hour. And yet, the task of modernizing legacy systems seldom stirs the animal spirits of today’s technology leader.

‍

But soft! What light through yonder backlog breaks? It is generative AI—and it is eyeing barnacled IT estates with great hostility.

‍

Generative AI has obliterated the “if it ain’t broke, don’t fix it” paradigm in IT, demanding that data not only be set free, but that it is accessible and LLM-able. Executives are actively dreaming up all sorts of new AI-powered revenue models and innovations: the last thing they want to hear is, “we can’t do that because it’s on a legacy system that no one has understood for years.”

‍

That’s why we celebrate when legacy systems are fully off the cloud. Yet, while migrating to the cloud is an important step, it remains to be seen whether simply being in the cloud removes the constraints of technical debt.

‍

As David Linthicum wrote recently, “Most workloads running on the cloud are unmodernized, and although they function, they are underoptimized. In many instances, they are sold to leadership as future projects where the application modernization would occur. Most of this just turned into technical debt, which means kicking the can down the road while paying for the road.”

‍

The can has hit a wall. And that wall is AI.

News & Views

McKinsey published a cogent view of how enterprises can best deploy AI. We particularly liked this: “competitive advantage comes from building organizational and technological capabilities to broadly innovate, deploy, and improve solutions at scale—in effect, rewiring the business for distributed digital and AI innovation.” Another reminder that AI is a technology, not a panacea—business design is the first principle.

‍

Legacy modernization projects are usually vast and unwieldy. Perhaps “micro-transformations” are the answer. It’s the opposite of the typical, big-bang projects, which have a high failure rate, since business requirements tend to keep moving over the course of 18-36 months.

‍

Think there’s just one seasonal peak? Think again. Tim Robertson, CEO of DHL Global Forwarding Americas, writes, “the idea of a single annual peak season could become irrelevant” and that logistics providers—and, we’d add, retailers—need to be prepared.

From the Orchard

The COVID-19 pandemic sparked more than just a love-hate relationship with surgical masks: it was one of the reasons Mechanical Orchard was created. Alex Konrad of Forbes reports.

‍

Mechanical Orchard CEO Rob Mee met with Block and File’s Chris Mellor over a cuppa’ and swapped storage stories. They’re likely also the first people to use the terms “jailbreak” and “mainframes” in a single sentence. Read the full article.

‍

Although Generative AI still can’t spell properly, it’s very good at many other things. But which ones? Mechanical Orchard’s Jeff Schomay digs in to discern which software engineering tasks are better suited to AI.

‍

Heading to GoogleNEXT? See you there. We'll be co-presenting with Google Cloud's engineering leaders at the session, “Mainframe modernization: Assess, refactor and rewrite using generative AI” (DEV-311). Get in touch if you'd like to meet up!

‍

Curious to learn more? Say hello@mechanical-orchard.com.

——————

Issue first published on April 2nd, 2024.