“The continuum is that which is divisible into indivisibles that are infinitely divisible.”
— Aristotle, from Physics

‍

‍

I flunked the first time I took my university’s introduction to philosophy course. The second time around, I aced it with this insight: if life and death bookend the concept of human existence, then really, it’s a continuum. Which means we’re both living and dying at the same time. Cue head exploding.

‍

Looking back, it was more adorable than profound, but it doesn’t make it any less true. And so it is with modernization.

‍

We are trying to fit what is essentially a continuous process into quarterly budgets, project dashboards, and system integrator contracts. This frames modernization as a discrete project that has a beginning and an end: it starts with a burst of investment followed by neglect—and a steady accretion of tomorrow’s legacy cliff.

‍

Instead, we should recast modernization as an operating discipline, alongside things like cybersecurity or compliance, areas that require ongoing attention—and funding—rather than episodic projects.

‍

As with my philosophical “insight", none of this is original: Gartner coined the term “continuous modernization” in 2018 specifically to counter the prevalent “big bang” approaches. It’s great in theory, but how do you actually make this happen? How do you take an unstructured language and convert it into a structured one? How do you take a batch system and move it to a real-time system?

‍

As with everything: one step at a time.

‍

First, rewrite the batch mainframe system into a modern cloud language one slice at a time, with the goal of replicating its exact behavior, except now in modern code: specifically, the existing behavior of the current system serves as the specification for your rewrite. This solves for moving from an unstructured language to a structured one (and eliminates disruption).

‍

Then, whenever you have a naturally cohesive group of workflows running, something you can treat as one “thing,” you can then rearchitect and add new things to it, make it a real-time, streaming system.

‍

In this way, you don’t have to wait for everything to be moved over before you can start making changes. You control the pace and the unit of system granularity that’s appropriate for your organization, in the workflow that delivers strategic change most quickly.

‍

The heart of this approach is the orchestration between the existing system and its modern counterpart. Because the old keeps running while slices are rewritten, you can experiment, validate, and adjust in real time—without putting the business at risk.

‍

That living bridge between old and new is what makes continuous modernization not just a project, but a way of being. (And it’s pretty darn hard to do, but we’ve figured it out.) Every system is, simultaneously, legacy and new. And the organizations that recognize and build to support this continuum will be the ones ready for whatever comes next.

‍

News and Views

MIT’s NANDA initiative reports that 95% of generative AI pilots at companies are failing. According to the report, the primary reason is the disconnect between the capability and the workflow specific to the enterprise. We suspect it’s also because old systems make it difficult to understand and reimagine those workflows, but “Companies surveyed were often hesitant to share failure rates.”

‍

Nevertheless, the shedding of people continues in the IT consulting sector. TCS announced the layoff of 12,000 people, while analysts estimate 400,000 to 500,000 professionals may be at risk over the next two to three years. Yikes.

‍

Fun stuff: part of moving forward is letting go. The soundscape to my adolescence is going softly into that good night. Goodbye, AOL dial-up. And hello robo-taxis. The Information reported that, in a survey of its (admittedly niche) readers, two-thirds have taken self-driving cars, and 99% of those were satisfied with their experience. (Subscription required.)

‍

From the Orchard

Modernization approaches are a hot mess of trade-offs. In Part II of our joint blog series with Thoughtworks, Shodhan Sheth and I write about how to balance future value and current risk. We’re calling it “test-first modernization,” but keen to hear whether that resonates.

‍

From time to time, we bring together a small circle of technology and business leaders for an evening that feels equal parts salon and supper (and sometimes art gallery!). We’ve scheduled two for London in mid-September and early December. With a cap of fourteen people each time, it’s a rare and rewarding way to exchange ideas. Reach out to request an invitation.

‍

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

‍

—

*Issue first published on August 26th, 2025. View all newsletters here

‍