Team Dynamics After AI (mechanicalsurvival.com)

At Agile Cambridge (2 Oct 2025) Duncan Brown — NHS England CTO and former head of the UK government AI incubator — argued that AI’s biggest impact on software teams is social, not just technical. He critiques the “Era of the Small Giant” narrative (small teams amplified tenfold by AI) by showing how rapid artefact production exposes, rather than solves, downstream social bottlenecks: governance, stakeholder expectations, publishing, and feedback. Prototypes become cheap but create coordination work; artefact throughput isn’t the true bottleneck. Brown frames AI as biased toward the legible and quantifiable, often at the expense of the “illegible” social and political work that actually determines system success. Technically, he contrasts two AI-enabled modes: amplification (boosting individual output) and hybridisation (tools that let one person temporarily hold multiple skill sets). Using UI design and GOV.UK-style design systems as an example, he warns against substituting designers with engineer+AI workflows: while AI can produce “good enough” components from a finite vocabulary, it converts designers’ read/write relationships into read-only outputs tied to tooling ecosystems, risking loss of deeper context, alignment and governance. The takeaway for the AI/ML community: build tools and metrics that surface and support social coordination, feedback loops and governance — not just faster artefact creation.