How do you build a machine that can win one of the oldest and most complex games in the world?
When Demis Hassabis – the founder of DeepMind – decided he wanted to beat the world champion at Go, his success was as much about the way he built his team as the technology itself.
Hassabis had created an organisation that allowed its researchers to spend a lot of time pursuing their own interests, but pulled them together at critical moments into multidisciplinary strike teams aimed at solving cross-cutting challenges.
By setting a clear goal, giving the teams explicit authority to go quickly and cross boundaries, Hassabis' AlphaGo was finally able to beat Lee Sedol, the world champion, in 2016.
In 2026 local government finds itself in a transformation bind. Many of us have been living hand-to-mouth since the destabilisation of the pandemic, balancing the next year without a long-term plan or eking out an exhausted strategy. The easy savings are long gone and so is the capacity to think about the hard ones. Where do we go next?
We should not be naïve about AI – it can improve productivity, but outside a few key sectors, is not yet replacing swathes of people. But we can learn from the organisations in the vanguard of this new technology.
I believe the next generation of councils – whether we are reforming from within or building afresh – will need to focus on three key operating systems.The first is the kind of problem-solving that Hassabis used to crack Go.
In Bath and North East Somerset (B&NES) we are approaching transformation with our own strike teams – multidisciplinary groups drawn from across the council to lead time-limited projects that reimagine our estate, transform our commissioning and build a new approach to neighbourhood working. By bringing our best people together with dedicated time to focus and license to drive forward at pace, we hope we can drive change without huge teams of consultants or programme managers.
We have set up teams in each of our services to work through their processes and identify where robots could speed things up, with the goal of releasing productive time rather than conventional savings. We also need to be ready for the moment when our residents start using their own bots to challenge council tax bands, manage planning applications and challenge parking tickets.
Critically, these groups operate within a reimagined budget process which is aimed at saving money through learning and change. We use a sprint methodology to bring our 50 most senior managers together once a year to identify the problems we want to solve in this year's budget cycle and reconvene monthly to check on progress.
The second operating system is the ability to use AI to drive much deeper automation of our business processes.
B&NES recently introduced an AI-native customer service platform which provides us with vast amounts of data about why people are calling us. But that is just scratching the surface. Look at the US insurer Lemonade, among the first of a new generation of AI native businesses which automates underwriting and approves simple claims entirely automatically. It takes a flat fee for each premium and donates a portion of any leftover money to charities chosen by its customers.
It is far from perfect – complex cases still need humans and they are hard to get hold of – but it is already possible to see services like planning going in much the same direction.
The next generation of automation clearly needs to go a lot further than just rolling out co-pilot licenses.
We have set up teams in each of our services to work through their processes and identify where robots could speed things up, with the goal of releasing productive time rather than conventional savings. We also need to be ready for the moment when our residents start using their own bots to challenge council tax bands, manage planning applications and challenge parking tickets.
This must all be in service of the third operating system – a shift away from delivering services to residents and towards generating value alongside our communities.
The greatest challenge facing local government is not processing transactions more efficiently, but helping places thrive in the face of rising demand, constrained resources and increasing complexity. AI can help us automate routine work and strike teams can help us solve difficult organisational problems, but neither is an end in itself. The prize is the capacity they create: more time to understand places, build relationships, convene partners and support communities to shape their own futures.
We have a lot to learn from the likes of DeepMind and Lemonade, but it is vital to remember that all the technology is in service to what really matters. Once the dust settles on local government reorganisation, the next generation of councils must fundamentally run on relationships.
Simon Parker is executive director of resources at Bath & North East Somerset Council
