Trends

The token bill is the cost everyone sees. Evaluation, human review, monitoring, and data maintenance usually cost more, and they are the lines missing from most business cases.

Prompts determine system behaviour, which makes them code. The practices catching up — version control, review, testing, ownership — are the ones every other config went through.

Most production tasks are narrow. Small models fine-tuned for them are faster, cheaper, and more predictable than frontier models, which is why they keep appearing in real systems.

Board conversations about AI are shifting from opportunity to accountability. The questions are specific, the answers require records, and most organisations cannot produce them.

Adding AI to an existing product is faster and produces a feature. Designing around what the model makes possible produces a different product, and the gap widens over time.

A policy saying humans review AI output produces rubber-stamping unless the interface, the workload, and the incentives support genuine review. Oversight is designed, not declared.

AI coding agents produce more code faster, which makes software quality a throughput problem for verification rather than a throughput problem for writing. This essay explains what changes when machines write code, why specifications and tests become the scarce input, how review must adapt, and what engineering leaders should change now to keep quality rising while output grows.

Banking is a data and process business under heavy regulation, which makes it both an ideal candidate for AI and one of the hardest places to deploy it. This essay looks at what changes in operations, risk, servicing, and product over the next five years, what moves to AI first, what stays human by design and by law, and how bank leaders should prepare without betting the franchise on unproven systems.

Consulting sells judgment delivered through people and documents, and AI automates much of the research, analysis, and document production that filled billable hours. This essay looks at what changes for consulting firms over the next five years, why the business model shifts from hours to outcomes and from slides to systems, what stays human, and how firms and their clients should prepare.

Customer support was the first function to adopt AI at scale and the first to learn that deflection is not resolution. This essay looks at what changes over the next five years as AI agents move from answering questions to resolving issues, how team structures and metrics change, what stays human, and how support leaders should prepare without repeating the chatbot mistakes of the last decade.

Data teams spent a decade building pipelines and answering questions, and AI now automates much of both. This essay looks at what changes for data engineering, analytics, and data science over the next five years, why data quality, semantics, and governance become the scarce inputs, how the team's role shifts toward data products and AI enablement, and how data leaders should prepare.

DevOps automated the path from code to production; AI now automates the work around that path, from infrastructure code and pipeline maintenance to incident triage and remediation. This essay looks at what changes for platform and operations teams over the next five years, which tasks agents take first, why guardrails and observability become the platform's core product, what stays human, and how leaders should prepare.

Finance teams run on documents, reconciliations, and cycles, which makes them a natural home for AI agents and a place where control failures are expensive. This essay looks at what changes for corporate finance over the next five years, from a monthly close toward continuous accounting, which work moves to digital FTEs first, how controls and audit evolve, what stays human, and how CFOs should prepare.

HR runs employee services, recruiting, talent management, and compliance, and AI changes each differently. This essay looks at what changes over the next five years, which HR work moves to digital FTEs first, why hiring and talent decisions are the most regulated frontier, how the employee experience shifts, what stays human, and how HR leaders should prepare while leading the workforce transition AI causes elsewhere.

Your agent depends on tool servers, prompt templates, and models you did not write. That is a supply chain, and it currently has none of the controls ordinary dependencies have.

AI systems are entering the scope of ordinary audit. The questions are mundane and the answers must come from records captured at the time, which most systems do not capture.

A blank chat box puts the burden on the user to know what to ask. Embedded AI that acts inside a known workflow removes that burden, and that is where adoption is going.

The first AI projects were handcrafted. The organisations shipping consistently now have templates, evaluation harnesses, and deployment patterns that make each project cheaper than the last.

The technology is the easy part. AI projects are converging on the shape of ERP implementations — process redesign, integration, data remediation, and organisational change.

Model choice gets the attention; data quality decides the outcome. Duplicates, stale records, and rules that live in people's heads cause most of the failures teams blame on the model.

Every AI system needs the same compliance controls, and building them per project is waste. They are consolidating into shared infrastructure the way authentication did.

Open-weight models are capable enough to be a serious option. Whether they are cheaper depends on utilisation, engineering cost, and whether the real driver is price or control.

Teams are not shrinking so much as changing shape. Specification, review, and operating probabilistic systems are growing; routine implementation is shrinking.

The first wave bought tools and hoped. The second embeds AI into specific workflows with measured outcomes, named owners, and operational funding. It is slower and it works.