As the year wraps up, the industry feels both more mature and more unsettled. AI is no longer “emerging.” Engineering leadership models are being re-examined. And teams everywhere are figuring out how to build systems that scale without losing their soul.
This month: a sharp look at the real job of a CTO, how engineering managers navigate the “pendulum,” and why prompting LLMs for five answers might be better than one.
Plus: Google’s lightning-fast image model, McKinsey’s State of AI 2025, an inside view of how M&S builds tech for millions — and a new Counter case study from East Midlands Airport.
Here’s December… Curated.
So… What Is a CTO?
Cory House offers one of the clearest breakdowns of what a modern CTO actually does — and what they should stop doing.
Less solution architecture.
More organisational clarity, technical strategy, and ruthless focus on leverage.
It’s a concise, grounded read, and pairs well with Majors’ take on the engineer–manager pendulum below.
The Engineer–Manager Pendulum
Charity Majors’ now-classic piece remains as relevant as ever. It reframes career progression not as a linear path but as a pendulum swinging between IC and management work — each informing the other.
Those who swing with intention tend to build healthier engineering cultures: more empathy, more psychological safety, and fewer brittle hierarchies.
Five Responses Are Better Than One
A new arXiv paper suggests a simple but powerful prompt pattern:
Ask an LLM for five responses — and probabilities for each.
The result?
Greater answer diversity without degrading quality.
It’s a small tweak, but it signals a broader shift toward steering models rather than relying on a single “best” output. Expect to see more structured prompting like this in 2026.
State of AI 2025
McKinsey’s annual report paints a picture of an industry in transition. Adoption is up. Expectations are higher. But many organisations have hit what McKinsey calls the “post-pilot plateau.”
AI isn’t experimental anymore — but integrating it into real workflows, governance structures, and operating models is where most teams are struggling.
Useful reading for any leader separating aspiration from actual capability.
How M&S Builds Tech for Millions
Hackajob’s recent deep dive shows how Marks & Spencer builds the systems that quietly handle huge scale and seasonal surges.
The takeaway: modern retail engineering is less about flashy interfaces and more about reliability, iteration, and deep empathy for how customers actually shop.
A refreshing, real-world look at impact over novelty.
Nano-Fast Image Generation
Google introduced Nano Banana Pro — a compact, hyper-optimised model capable of generating images at near-instant speeds.
It’s tailored for consumer devices rather than data centres, hinting at a future where high-quality image creation is native, offline, and almost frictionless.
One to watch as edge-AI moves from novelty to expectation.
Case Study: East Midlands Airport
Counter partnered with East Midlands Airport to accelerate their digital transformation — modernising legacy processes, improving operational visibility, and giving teams clearer, real-time data to work with.
It’s a practical example of how targeted engineering can reshape how an entire operation runs day to day.
Want more like this?
Want Counter’s Curated news sent straight to your inbox? Pop your email in here.
