#16 Doing less, the hard way

Government design principle number two is "Do less". It's easily overlooked, but it might be the most powerful. Principle four is "Do the hard work to make it simple." Put them together, and you get a nice paradox: sometimes doing less means doing more work upfront so everyone else doesn't have to.

This feels especially relevant now that AI is accelerating how quickly teams can build digital services. Speed is great, but without shared patterns, you get inconsistency at scale. A component library gives teams (whether they're using AI tools or not) something consistent to build from.

We’ve been working on a client project that required rapid prototyping using the GOV.UK Design System (GDS) – the standard set of styles and components used across government services. The client wanted it built in React, but the existing open source React component library for the Design System hadn't been updated in two years, in which time the styles of the GOV.UK styles had been updated significantly. So I built a new one and open-sourced it.

The approach was deliberately simple: create reusable components that lean on the existing GDS styles rather than maintaining a separate copy. The result is a library with familiar, standardised patterns and accessibility – and less code, so it’s simpler to maintain over time. During development, I also found and responsibly disclosed a security vulnerability in Storybook (a widely used development tool).

At IF, we’re strong supporters of creating repeatable solutions for common problems and making things open (see our design patterns catalogue). So it was a no-brainer to make this available openly to other teams across government and beyond.

For 10 years, IF has been helping large organisations build customer-facing services that scale safely, earn trust from the start and deliver long-term impact. We prototype, test, and launch AI products and services that people believe in and want to adopt, while helping organisations change the way they work in the AI age.

Learn more or talk to us.

What we’ve been working on

Scaling AI prototypes for cities

We're in the technical phase with cities on the Bloomberg Philanthropies City Data Alliance, helping them make their functional AI prototypes work with real data. For some cities, the next step is maintaining trust through community engagement; for others, it's achieving greater transparency by making the technology more explainable to the people using it.

Supporting the safe use of public sector data

We're working with The National Archives to improve the Open Government License guidance to support the safe and correct use and re-use of Public Sector Information. We’re looking for user research participants among public sector information providers and people who re-use this data.

What we’ve been reading

Computer says ‘maybe’. Gone are the days when the computer says no. GenAI replaces that certainty with probability, with little accountability. Tom Wynne-Morgan argues that the hard work isn't training better models – it's deciding who carries the risk when the machine says "maybe". Read the article on Medium.

AI and the end of friction as a policy lever.  Many services in the public and private sectors rely on bad user experiences to manage demand. Tom Loosemore highlights why AI agents are about to make that impossible. Read the article.

Design just got serious again. AI can generate the output, but it can't provide the judgement. A piece on why taste, systems thinking, and editorial instinct now matter more. Read the article on Substack. 

AI doesn’t reduce work – it intensifies it. Organisations need to adopt an AI practice that structures how AI is used and sets boundaries around how work is expanding in response to new capabilities. Read HBR’s article.

I'd love to hear how you find our React component library and if you're using it in your work – drop me a message at [email protected].

Until next time,

— Matt and the IF Team

This month’s edition was written by Matt Gill, a full-stack engineer specialising in AI learning and a member of IF’s network. 

Find us on LinkedIn and Medium for more insights and updates.