“Own the customer's journey, or rent out the rails. There is no comfortable middle.”
‍
– Dharmesh Mistry

‍
Most technology cycles evolve slowly, giving incumbents time to adapt and benefit from them — the PC, the internet, mobile, even AI's first wave. The smartphone was the exception: GPS, banking, social media, and a browser converged into one device, and Nokia and Motorola were gone within a few years.

Dharmesh Mistry has spent four decades building banking technology — at Lloyds, NatWest, and as chief digital officer of Temenos, where he had a hand in the UK's first mobile banking app. His forthcoming book, The Intelligent Ledger (with Lou Carlozo, Palgrave Macmillan, available for pre-order), argues banking is nearing its own version of a smartphone moment. I sat down with him ahead of the book's release.

—

Tell me about the book. What’s it about, and who is it for?

It's written for a general audience but aimed at banks. Four things are converging (aka “Quanvergence”): agentic AI, quantum simulation, digital twins, and tokenization — which completely disrupts how money moves, and lays bare the fiction that banks “own the customer”.

‍Tokenization means settlement can happen in real time. What does that actually do to a bank that's been running on batch processing for decades?

It starts with agreeing what "the core" actually means. Most people would say it's just a ledger, with everything else — credit risk, AML, KYC — sitting around it as an extension. Tokenization turns that ledger into something shared: one record instead of every bank keeping and reconciling its own copy, collapsing the old gap between recording a transaction and settling it. That's a hollowing out, not a wholesale replacement: banks get real-time settlement working first, then migrate the systems around it. Whether that's on a mainframe matters less than whether reconciliation already runs in real time, since most mainframes were built for batch and haven't all shifted to a realtime window. Banks with real-time core banking already in place have the luxury of time to adjust to the new possibilities.

What does all this mean for mainframes?

Two things actually matter: real time, and scale. Neither of these is about the hardware: cloud infrastructure and mainframes ultimately run on similar hardware. But cheap, instant transfer removes the reason to batch payments monthly, so systems built for millions of transactions now need to handle billions, as agents make decisions at machine speed. What changes is throughput: card networks handle something like 10,000 to 15,000 transactions a second today, and if that needs to become a million, the constraint is less the machine than whether the software running on it can keep up.

Given all that, what should an IT leader actually do first?

Read the book. Every chapter ends with homework — questions a board should be asking itself — and there are frameworks for the harder decisions underneath: whether a bank chooses to be a platform, owning the customer relationship, or a pipe, manufacturing product for someone else's platform. You can work through it chapter by chapter, using your own organization’s characteristics.

‍
You can get a free copy of Dharmesh’s book! Or, if “quanvergence” is not your cup of tea, you can choose from other books in Mechanical Orchard’s first book giveaway, in honor of National Book Month (October).

‍

News and Views

You know that feeling when something you’ve said over and over again is finally being validated by “name brand” consultancies? We felt that way reading BCG’s Six Myths CIOs Must Avoid in AI-Based IT Modernization. I mean, look at Myth 3: AI Can Explain Everything from the Code Base. We’ve written extensively (here and here) about how business rules extraction is important but insufficient. Then there’s Myth 5: AI-Assisted Coding Is Enough to Accelerate IT Modernization. See this and this. And finally, the humdinger: Myth 6: AI Reduces the Need for Testing and De-risks Cutover. Hello, white paper.

‍Ensono dropped their latest State of IT Modernization survey and we gave yet another nod of recognition. A few tidbits: 71% of respondents said modernization initiatives have exceeded their original budgets, with 27% exceeding budgets by 51% or more. Difficulty integrating AI into existing workflows or business processes is the top-cited barrier to achieving expected ROI from AI. Yes, yes, indeed.

‍Addy Osmani's latest draws a clean line between brownfield and greenfield engineering: agents can move fast where the codebase is the whole story, but most real systems hide constraints outside the tree. Mainframes are the extreme case (his “red zone”), where the people who knew the "why" are often gone along with the documentation. While his assertion that BREs are helpful but insufficient echoes ours — “write down what the code can't say” — we’d go beyond and posit that the source of truth is the system behavior.

‍

From the Orchard

We love our readers, and we love reading, so it’s only natural we celebrate this upcoming National Book Month with a giveaway: from the wondrous world of code to business transformation advice, take your pick from our bookshelf.

Mechanical Orchard is now ISO/IEC 27001:2022 certified. Schellman, an independent, ANAB-accredited body, audited our Information Security Management System and confirmed our security controls operate as described, across the Imogen platform, our DevOps environments, and how we handle customer data during an engagement. Our CISO Sam Pierson explains what the certification covers here.

‍We’re sponsoring Carahsoft’s networking reception next month at the Google Public Sector Summit 2026. If you’re attending on October 20th, come by Glass House at Dirty Habit from 5:15 pm for drinks, nibbles, and a chat about your modernization strategy.

‍Finding the budget for shiny new things is easier than proving their impact afterward. At the New York edition of the ISG AI Impact Summit on October 29th at 2pm, our COO Edward Hieatt will take to the stage to argue how to move faster and more safely by putting verification first. Register here with our referral code, ORCHARDVIP.

‍

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

‍

—

View all newsletters here

‍