Integrating legacy systems with new tools

The right question isn't "rewrite or not"

Faced with an old ERP that still runs the operation, the first instinct is usually binary: rewrite everything or keep suffering. In practice, most of the gain comes from a third option — integrating the legacy system with new tools on top, without touching the core that already works.

Rewriting is expensive, slow, and risky precisely in the part of the business that can't stop. Integrating is incremental: it connects what already exists to a new layer, without betting the entire operation on a months-long project.

Real examples of incremental integration

Deloitte documents concrete examples of this path. Amazon used generative AI to speed up Java application upgrades, cutting upgrade time from 50 developer days to a few hours — with an estimated $260 million in annualized efficiency gains. Airbnb updated React components in six weeks, a process a manual estimate had pegged at a year and a half. Goldman Sachs improved developer proficiency by 20% using auto-coding tools on top of its existing systems.

In none of these cases was the legacy system replaced from scratch. Modernization happened in the integration and automation layer built on top of what already existed.

What to evaluate before integrating

Three questions guide the decision: does the legacy system still deliver the expected business outcome (even if slowly or manually)? Is the pain point the technology itself or the lack of connection to modern tools? And is there an API, even a limited one, to serve as a bridge — or would integration require building that bridge from scratch?

When the answer to the second question is "lack of connection", integrating tends to solve the real problem faster and with less risk than rewriting.

When rewriting is still the right answer

Incremental integration doesn't fit every case. When the legacy system no longer runs on any supported infrastructure, when no one understands the code anymore, or when the cost of maintaining it exceeds the cost of replacing it, rewriting stops being overkill and becomes necessary. The difference is deciding this with data, not generic discomfort about old systems.

Where Diglion comes in

Diglion maps where integration solves the problem with less risk and where a rewrite is actually justified, before committing budget to either path.

Sources consulted

• Deloitte Insights, Legacy system modernization, retrieved 2026-08-11.