Apple Platform Engineering

In 2025 IBM produced a report that showed Mac computers are cheaper and safer to maintain in enterprise environments, costing around $500 less per device over its lifetime.

This came at a particularly bad time for Microsoft, who have gambled everything on hyperscale AI, injecting ads into everything, and building some of the most poorly considered surveillance tools into their operating systems.

This reality has yet to hit most organisations, who will soon be making a decision between Linux and Apple for their user devices. While we love working with Linux, it’s not mature enough for regular users desktops just yet. We believe the smart play is to focus on Mac; yet almost nobody is doing disciplined, scalable engineering on Apple platforms.

But why Mac? Sure their devices are easier to use and have a performance and efficiency profile no other machines can touch, but beyond that their Silicon is nothing short of revolutionary.

We’ve been working with MLX, the framework that allows among other things running machine learning models locally in a very efficient way, since it was announced. We’ve open sourced MLXPromptCache, bringing sub 2 second Time to First Token for our CyberBench local inference driven security analysis platform.

Our mission is to lead the way in modern Mac based productivity and business applications. We spent decades bringing rigor to enterprise cloud. We are bringing the same rigor to the Mac ecosystem!

Built to scale

Most Mac applications are built as one thing, by hand crafted with love. This is fine right up until you need a second implementation, or a headless variant, or the same core running somewhere it was never designed to run.

We build on the Hexagonal architecture pattern, so the core of an application knows nothing about how it is being driven. That is not an academic preference. It is the reason our desktop application became available on the 1st of July, we were able to begin the mobile version on the 2nd, and it passed Beta Review on the 5th. The same core capabilities wrapped in a tool for a different user use case.

The same pattern gives us a headless HTTP variant that has survived twelve hour soak and storm tests, which is the sort of thing you only get to find out about if the architecture let you try it in the first place.

Are Macs the next data centre appliance?

We think so, the headless variant of the Tanuki local voice to doc, inference driven application proves this. We built this variant to sell directly to businesses interested in using Macs as a network appliance, and the performance that we have seen is shocking. The whitepaper linked below goes into the details, but we found a low-end commodity Mac running modern swift applications can approach the performance of Go under load, without any run away heat issues.

We’ve proven that a network connected Mac Mini can run inference for an entire office without breaking a sweat. Go stand next to a SambaNova rack and see what it does.

Delivery discipline, where it has barely arrived

Signing, notarization, provisioning profiles, reproducible builds, runner fleets and TestFlight promotion are still largely manual across the Apple ecosystem. Most teams shipping on Apple platforms have quietly built their own fragile version of the same pipeline, and more often than not it lives in one person’s head.

This is the work we have been doing in Energy, Banking and Cyber-Security for years, pointed somewhere new. Our Mac and iOS apps are initiated from Backstage scaffolder templates, with testing and deployment affordances built in. We deploy HelloWorld to test flight minutes after initiating a new project repo.

We can help you get your Apple delivery into a pipeline your team trusts, with the same auditability and the same ability to revert to a known good state that you would expect of anything running in your cloud.

What is on your machine is yours

Apple-silicon has changed what is possible locally, and to us it is the purest expression of how we think about software. Inference that runs on the machine has no egress to manage, no third party holding your data, and nothing phoning home.

Any piece of software we build starts here, network access is only for things the user requires, not providing an extra revenue stream for us or one of our providers. Our relationship is with you our customer, and we are not interested in selling you to the highest bidder.

We are further ahead on MLX than most, and our MLXPromptCache module is already open source, so you can judge the standard of the work before you engage us on any of it.

App Store Readiness

In 2025 the Apple App Store rejected over 2 million problematic apps. This is a pretty huge number, and it’s hard to quantify just how difficult it must be to verify at this kind of scale. It is also one of many data points that help explain why so many first time submitters have such enormous difficulty publishing their apps.

Once the technical issues have been ironed out, then we come to the softer aspects of the application. Rejections come for many reasons, and often the feedback from Apple is opaque and the instructions sparse.

In March 2026 we decided to publish our first tool to reach the App Store, Tanuki desktop. We were proud that it became available on the 1st of July, and given our use of the Hexagonal architecture pattern, we were able to begin the mobile version of the app on the 2nd of July, with it passing Beta Review on the 5th of July.

Whether you are preparing for a future release and need assistance with your strategy, or you have been stuck in review for a few weeks, we can help you get through it.

Where this is going

The Mac is going to carry a great deal more serious software than it does today, and the organisations deploying it will want the same assurances they already expect everywhere else. Provenance for what they are running. Pipelines that do not depend on one person. Inference that stays on the machine.

That work is barely started, and we intend to lead it.

Decades of enterprise delivery discipline across Energy, Banking and Cyber-Security, and today the same discipline applied to Apple platforms.