Insights
On AI, engineering and building software that lasts
Trends, field notes and guides on AI, machine learning and software engineering: what is changing, what works in production, and what we would do differently.
Trends
What is changing in AI and engineering
Analysis of where the industry is moving on AI, machine learning, agents and delivery, and what it means for teams building real products.
Trend · August 25, 2026
Agentic AI is moving to production, and it breaks the old playbook
The industry moved from single-shot prompts to agents that plan, call tools and act over many steps. That shift changes what you build around the model.
ReadTrend · August 18, 2026
Context engineering is the new prompt engineering
As models got better at following instructions, the hard problem moved: deciding what goes into the context window, in what order, on every call.
ReadTrend · July 15, 2026
Small, specialised models are quietly winning
The reflex is to reach for the biggest frontier model. For many production tasks a smaller open-weight model, fine-tuned, is faster, cheaper and good enough.
ReadTrend · June 30, 2026
AI coding agents and what they do to a software team
Coding assistance moved from autocomplete to agents that open pull requests. The productivity story is real, and so is the shift in where the bottleneck sits.
ReadTrend · June 9, 2026
MCP and the standardisation of AI tool use
The Model Context Protocol has become the common way to connect AI applications to tools and data. Here is why a standard connector layer matters and where it helps.
ReadTrend · May 19, 2026
Securing LLM applications is now its own discipline
Prompt injection, tool-based data exfiltration, agent confused-deputy attacks. The threat model for AI features is specific and the defences are now standard.
ReadField notes
From live engagements
Short write ups from real projects: a decision, a trade off, a thing that broke and what we changed.
Field note · August 12, 2026
Handover is a feature, not a phase
The value of a build is only realised if your team can maintain it after we leave. We treat that as something to design, not a document to write at the end.
ReadField note · July 29, 2026
How we scope fixed-price work so the estimate survives contact
Fixed-price software goes wrong when the estimate is a single number produced before anyone understands the problem. Here is the process we use instead.
ReadField note · June 17, 2026
Introducing trunk-based delivery to a team that had never tried it
Long lived branches were the norm, releases were monthly, and every merge was an event. Here is how we moved to trunk-based delivery without a big bang cutover.
ReadField note · May 6, 2026
We deploy a walking skeleton before we write a feature
The first thing we put in production is a thin end to end slice that does almost nothing. Here is why that pays for itself in the first week.
ReadGuides
Guides and playbooks
The checklists and reference material we use internally, cleaned up and shared.
Guide · April 22, 2026
Evaluating an AI feature before you ship it
A prototype that looks good in a demo tells you almost nothing about production behaviour. Here is how to build an evaluation that does.
ReadGuide · March 10, 2026
A launch readiness checklist for production software
The checklist we run before putting anything in front of real users: reliability, security, observability, support and the rollback you hope not to need.
ReadGuide · November 18, 2025
Adopting a design system without stopping feature work
You rarely get to pause the roadmap to build a design system. Here is how to introduce one incrementally, paying for it with the work you were doing anyway.
ReadReports
Engineering reports
Original research on how teams build software, based on surveys and our own delivery data.
Events
Webinars and events
Occasional sessions on the topics above, with time for questions. No pitch.
We run these a few times a year and announce them to the newsletter first. Subscribe below to hear about the next one.
Have a problem worth writing about?
Tell us what you're working on. We come back within two business days with a point of view and next steps.