“Nothing in life is to be feared, it is only to be understood.”
‍
– Marie Curie

‍
I learned something new today. It’s called the Overton window, and it describes how, in politics, only a narrow range of policy positions on an issue are considered “acceptable” for a politician to hold without being seen as radical or unelectable (think: same-sex marriage). But this range, or window, slides in pendulous ways and doesn’t always result in a concrete shift (think: DEI programs).

Digital transformation is similarly pendulous - think of how it took a worldwide pandemic to normalize remote working, apoplectic CEOs aside. Cloud storage of sensitive data used to be unthinkable: now it’s common. In both cases, it took a few goes before advances became commonplace - advances that then opened the door to more innovations.

We see this with mainframe modernization, too. Last February, the pendulum seemingly swung to “AI can do all the COBOL things” (Anthropic). It quickly swung back to, “70% of mainframe modernization efforts using AI will fail” (Gartner). Meanwhile, no one is disputing that vast swathes of the existing mainframe estate represent real constraints on what can be done to meaningfully take advantage of AI: the question is whether it’s even worth trying. Sensibly, caution here is the correct response to a system where failure is expensive and hard to reverse.

Yet, every dollar of AI ROI an enterprise is chasing runs through that brittle, Winchester Mystery House of a system, the one that keeps the vital organizational organs pulsing and alive. The mainframe hasn’t become less of a constraint because the industry wishes it weren’t so, or because business rules can now get extracted from code. To truly innovate on a living, breathing system requires rewriting its behavior — how it integrates with the rest of the systems, what it does about rounding, encodes data, handles runtime errors, etc. — into more malleable forms.

So what does it take to actually shift sentiment from ennui to hope, from wishful thinking to action? Personal contact.

Research found that the “contact hypothesis” — when a personal experience turned an abstraction into a person — was the most significant reason the Overton window shifted on the topic of same-sex marriage. Minds changed when relatives, friends, coworkers came out because it’s hard to “other” someone you love and/or respect.

Mainframe modernization needs its own version of contact. Not another vendor statistic about how fast a bajillion lines of COBOL were converted into Java, and not another analyst report projecting failure rates across an industry. Proof has to be applicable to their system, specifically, a specific, undeniable case that displaces a specific, previously abstract fear.

That’s what our proof of concept is for. Not a demo running someone else’s sanitized sample data, not a slide deck extrapolating from a different company’s mainframe. We’ll run Imogen against a real slice of your codebase, in days rather than months, producing a Java replica characterization tested against your own logic. We’ll do it at no cost, under NDA, and your code never leaves an MO-controlled instance. You get an actual modernization roadmap and a demo environment your own technical leads can put their hands on — and find out it’s not so scary and impossible after all. And that’s how the window moves: one system at a time.

‍

News and Views

Gartner published its first Magic Quadrant for AI-Augmented Code Modernization Tools and gave Mechanical Orchard an honorable mention. The report cites mechanisms for functional equivalence, auditability, and transparent, verified AI outputs as non-negotiables for vendors in the industry, predicting that modernization delivery will shift toward tool-led, outcome-based engagements.

Last month, UK customers at Lloyds, Barclays, Halifax, HSBC, and Monzo received failure messages on transfers that had actually settled, with some initiating payment twice before the banks began telling people to check their accounts. It’s interesting that at least for one afternoon, major financial institutions themselves were telling their own customers: don’t believe what the system says, look at what it actually did.

‍

From the Orchard

Dark factories can prompt fears about who has control, but our VP of Strategy, Rachit Awasthi, sheds light in a multi-part series on how, even with the lights off, you can have even more visibility than the opaque processes people have accepted as the status quo. Part 2 likens the design of our software factory to a theory from cognitive science, and explains why the comparison is good news for anyone maintaining oversight.

While they might not be specific to your system, we do have customer case studies in case you don’t want to be first. We’ve detailed what we achieved for a F500 global manufacturer and presented on stage with our customer, SulAmérica at Google Cloud Next. But don’t take our word for it — try Imogen for free.

Mechanical Orchard will be at the ISG AI Impact Summit, and our COO, Edward Hieatt, will be speaking October 29th at 2 pm. Come join us in New York City: redeem your complimentary registration with our guest code, ORCHARDVIP.

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

‍

—

View all newsletters here