“Generative AI Engineer” was the UK’s single fastest-growing tech job title through 2025, and AI-adjacent roles — AI governance manager, responsible AI lead, AI ethicist, automation practitioner — are filling out job boards through 2026. It would be easy to read that as “every software team now needs a dedicated AI specialist.” The hiring data says the opposite for most commissioning decisions: AI engineering roles still make up a small slice of the wider tech market, and most organisations continue to prioritise full-stack software engineers who use AI coding tools well to work faster, rather than hiring narrow specialists to build or fine-tune AI models in-house. Commentary on the UK startup market this month put the underlying mood plainly: “AI still attracts attention, but buyers and investors now want proof, not polish.”
Why the distinction matters for a build, not just a hire
Building most AI-powered features — a chat interface, a recommendation engine, an AI-assisted workflow inside an existing app — very rarely requires a team that trains its own models. It requires a team that integrates well-chosen third-party models (Claude, GPT, Gemini and the rest) competently, uses AI coding tools to move faster on the surrounding engineering, and applies the same architecture and security discipline it always has. That’s a different, and much more common, skill profile than the “AI engineer” title implies, and it maps far more closely onto a strong general software team using modern tooling than onto a specialist ML hire.
Why this matters if you’re commissioning software
If a project brief or a vendor’s pitch leans heavily on “we have AI engineers,” it’s worth asking what that actually means for your specific build: are they training models, or are they building software that calls existing ones well? For the vast majority of commercial AI features, the second is what actually gets shipped, tested, and maintained — and it’s a capability a good AI-assisted development team already has.
So what
Before writing a job spec or a project brief around “needing AI expertise,” get specific about whether the work is model development or AI-assisted product engineering — they’re solved by very different teams. See our AI-assisted development approach for how we scope AI-powered features without over-specifying the team, or get in touch to talk through what your build actually needs.