Blog
How we build real systems.
Short posts on the decisions that matter in production. AI, modernisation, platform engineering, and the problems that show up once a project is past the demo stage.
All posts
Field notes from the work.
Modernisation · May 2026
The SAP Migration Mistake Most Teams Make First
Teams start by moving code. The right move is starting with the data. Here is the sequence we use on every SAP migration and why getting this order wrong causes the big-bang weekend nobody survives cleanly.
AI Engineering · May 2026
How We Decide Whether to Build a Feature With AI or Without It
AI is not always the answer. We use a short decision framework on every new feature request that has kept us from wasting six figures on models that a rule-based system would have handled better. The two questions are simpler than you think.
Platform Engineering · April 2026
Platform Engineering Is a Product Problem, Not an Infrastructure Problem
Most internal platforms fail because the team building them measures success by uptime and deployment speed, not adoption. Here is how we think about internal platforms as products that engineers actually choose to use rather than tolerate.
Manufacturing AI · April 2026
What Predictive Maintenance Actually Needs to Work on a Factory Floor
Sensor data alone is not enough. Good failure prediction needs labelled downtime events, clean master data, and an ML pipeline that a maintenance engineer understands. Most pilots skip at least two of those and then wonder why the model stops predicting anything useful after month three.
Cloud Engineering · March 2026
Cloud Cost Is an Architecture Decision, Not a Finance Problem
By the time the bill arrives, the decisions that caused it are already in production. Over-provisioned databases, synchronous calls where async would do, logs nobody reads: these are fixable at design time. We walk through the five patterns that show up on every high-cost cloud project.
GreenOps · March 2026
What Carbon Accounting Actually Looks Like in Code
Emissions reporting for cloud workloads is mostly guesswork unless you instrument your systems correctly. The cloud providers' built-in tools are not enough for a real audit. Here is the measurement stack we use and the gaps you will find in every major provider's dashboard.
Systems Integration · February 2026
Why Most Integration Projects Fail in Month Four, Not Month One
The first three months look fine: the happy paths work, demos go well. Month four is when the edge cases that were waved away in the requirements start arriving in production. Here is the integration testing approach we use to surface them before that happens.
Healthcare AI · February 2026
Clinical AI Without the Liability: Testing Patterns That Satisfy Regulators
AI in clinical settings needs audit trails that work for patient safety reviews, not just unit tests. We cover the testing patterns, safety guards, and documentation approaches that have passed scrutiny across three healthcare projects in the UK and US.
Product Strategy · January 2026
The Three Questions We Ask Before Starting Any Build
Most projects start with a solution and work backwards to a problem. We start with three questions that take twenty minutes to answer and save months of building the wrong thing. If your team cannot answer all three in a room together, the backlog is not your biggest problem.
Agile Delivery · January 2026
Why We Stopped Estimating in Story Points
Story points measure effort. What delivery actually needs is cycle time: how long does a piece of work take from start to done? After moving a dozen teams from points to flow metrics, the answer to "when will it be ready?" became a number instead of a shrug.
Want a post on a specific topic? contact@mverve.com
Let's talk
Ready to engineer your industrial future?
Tell us what you're building and we'll tell you, honestly, how we'd approach it. We read every message and reply within one business day.