What's actually different now
Twelve months ago, AI coding assistants were useful but limited — good for autocomplete, occasionally helpful for boilerplate code, but unreliable for anything complex. That's changed significantly. Current AI coding tools can generate complete, functional components from a description, write test suites, identify bugs, suggest performance improvements, and convert designs into working code. They're not perfect, but they're fast enough and capable enough that a developer using them well can do in a day what previously took three.
For clients commissioning web projects, this matters. Not because the work is being handed to a machine, but because the economics of development are shifting. Tasks that used to consume a senior developer's time — writing documentation, generating standard CRUD interfaces, building data validation logic — are now faster. That time goes back into the parts of the project that genuinely require human judgment: the architecture decisions, the UX thinking, the edge cases that no AI will anticipate because they require understanding of the specific business.
What this means for cost and timelines
For straightforward projects — brochure websites, standard e-commerce, content-driven sites — AI-assisted builds are meaningfully faster and cheaper than traditional development. We're not talking about a 5% saving. We're talking about builds that took 6-8 weeks now taking 3-4, and the cost coming down accordingly. This is genuine, not marketing.
For complex custom applications — bespoke booking systems, multi-tenant SaaS platforms, custom integrations with unusual APIs — AI speeds up the component-level work but doesn't change the project structure significantly. The complex parts of a complex project are complex for reasons that have nothing to do with typing speed. Here, AI helps with the surrounding work: test coverage, documentation, data migration scripts, deployment configuration.
Why a human still needs to be in the loop
AI-generated code is only as good as the person reviewing it. The current generation of AI coding tools makes confident mistakes. They'll generate code that looks correct, runs without errors, and passes obvious tests — but fails in edge cases, has subtle security vulnerabilities, or works fine in isolation but breaks when integrated with a real production system. A developer who trusts AI output without reviewing it is going to have a bad time, and so are their clients.
We use AI throughout our builds. Every line it generates is reviewed, tested, and understood before it goes into your project. We don't use AI as a replacement for thinking about your project — we use it to think faster about the parts of your project that don't require novel thinking. The part that requires understanding your business, your users, and your specific constraints is still done by humans who have had actual conversations with you.
What about AI-generated design and content?
AI image generation has made certain kinds of placeholder and stock imagery obsolete. AI copywriting tools can draft initial content quickly. Both are useful starting points and both require human refinement before they're appropriate for a business website. Generic AI copy reads like generic AI copy — your visitors have seen enough of it to notice. Generic AI imagery tends toward a specific aesthetic that dates quickly.
We use AI tools for initial drafts and to speed up the iteration process, but the content that goes live on your site is edited, refined, and approved by a human who understands your brand and your customers. The goal is a website that sounds like you, not like a language model's approximation of a business.
What to expect from an AI-assisted build
If you're commissioning a website or app from us, here's what changes and what doesn't. The timeline is typically faster. The cost is typically lower than a traditional development agency. The quality bar is the same — we test thoroughly, we optimise for performance, we don't cut corners on security. What you get is a project that benefits from the efficiency gains of AI tooling without the risks of an AI-first approach where nobody's actually checking the output.
You can read more about our website service and our app development service to understand what a project with us involves.