Today’s been gloriously sunny, cold and crisp. A good autumn day with winter on the way.

Spent a huge portion of the weekend in the garden, building raised beds and sowing broad beans, after getting some inspiration in South Africa last week. We visited Babylonstoren, a hotel with a vineyard, working farm and farm-to-fork restaurant, situated near Paarl in the Winelands. I’ve always wanted a kitchen garden but never put the effort in to grow veg – so I’m starting now, new hobby ahoy.

Also picked up a book on permaculture by Graham Burnett, which is inspirational. Of course I’ve watched loads of videos on YouTube too, and it seems like a much better way of growing.

For the last few months I’ve been thinking I really need a slow hobby, something physical where it takes a while to see the fruits of your labour. Running counts, but I wanted time to make things too. Another bonus with gardening is that it makes use of our garden, which I neglected this year. (My partner did well to grow things, but all I mustered were some wildflowers.)

Promise this won’t turn into a garden blog. On to the work stuff, which is why you’re probably here.

As mentioned, I was away last week, but the team conducted the first benchmarking session, testing the AI-powered prototype alongside manual processes. The testing rolled into the start of this week too, and we’ve shared early insights from users.

Time is tight and I’m really hoping we can line people up for testing and research over the next two weeks, but you know what it’s like in the run up to Christmas. We’ll see. Fingers crossed.

I’d really like to do a diary study with local planning authorities, leaving them with the tool for a week to see what they get up to. We could review the diary together and co-design new features quickly. But the most important learning would be whether they achieved the outcome or not. We’re in a novel space so it’s really important to see them use the tool, use it often, and get their data into back-office systems. That’s the main thing that matters.

A new design-dev cycle

The team are letting me trial a new method for design and developing features rapidly. It’s not wholly new, it takes inspiration from various places, but the main goals are:

1. eradicate the hand-off between designers and developers
2. produce as many options as possible, and
3. embrace technical and design constraints as early possible.

We’ll have two designer-developer pairs working on rapid prototypes inside a few days. They’re welcome to use AI tools if they want – coding agents are really quite good at backend and middleware mechanics, but I’m yet to be impressed by their frontend and design chops.

People are talking about how AI flips software development. Previously the process to build something often took longer than the process to design it, but these coding agents change that. Engineers can pull something together really fast, and design has to do the UX thinking and ‘clean up’ afterwards.

I hate the sound of that. I also wanted to bring a social element to the work, finding better solutions through dialectic and dialogic conversation. Opposite views can be synthesised into a better whole, and discussions which do not reach a common ground can still improve the participants’ understanding of each other.

A quote from Richard Sennett was inspirational:

“De-skilling is occurring in the social realm in equal measure: people are losing the skills to deal with intractable differences as material inequality isolates them, short-term labour makes their social contacts more superficial and activates anxiety about the Other. We are losing the skills of cooperation needed to make a complex society work.”

It’s important to centre the value of human cooperation in our craft, otherwise we’ll atomise each role into a human-computer collaboration, losing the social bonds that make teamwork work.

There was another quote from a talk by someone at Perplexity that struck a chord too:

“The future isn’t about AI replacing developers and designers—it’s about freeing them from routine work to focus on architecture, strategy, and creative problem-solving that machines cannot yet replicate.”

I’ll share more in January once we’ve trialled it a little bit. Keep your eyes peeled.

My next posting

Even though I’ve still got four months left on this contract, I’m starting to think about what I’d like to do next. It feels sensible to take some time to learn new skills, like going deeper on either design or development. Doing some data engineering might be fun too.

I’d like to try a different domain too. Perhaps something private-sector, or a government that isn’t the UK.

A 3-day-a-week contract might be good for building up another arm of the business.

Bookmarks

• Every organisation has some madness, 3 mins. Fantastic piece by Katherine Wastell. Inspired me to start writing a thing about leadership I’ve been noodling on for a while. Real change needs clear direction, behavioural shifts, and balanced thinking plus doing.
• The Real Reason Agile Created Chaos in Product Management, 4 mins. Agile itself is fine, but many companies ruined product management by turning strategic product managers into tactical product owners. Backlog jockeys. This caused lost market focus, role confusion, and burnout.
• It’s time to get serious about design, 7 mins. A primer on true design, not treating design as making small fixes or interface tweaks. Designers must have freedom to explore, create, and solve complex problems beyond simple tasks. Leadership should create space for creativity and empower designers to drive real change.
• Crossing The AI Divide, 3 mins
• ‘It fully changed my life!’ How young rewilders transformed a farm – and began a movement, 12 mins
• “There is no magic” – An hour with Public Digital’s Katherine Wastell, 7 mins
• Social media is dead — none of my friends are posting any more, 4 mins
• Olmo 3 is a fully open LLM, 6 mins
• The health record keeps the score, 3 mins
• The disciplines theory of government, 21 mins
• Unordered Note 14/11/25, 3 mins
• Talking AI With Martin Wolf, 30 mins
• 43 minutes per staff member per day, 4 mins
• From War Rooms to Obeya, 7 mins
• Output or Outcome: Which Drives Your Team?, 2 mins
• Digital identity and the UK government’s announceability problem, 6 mins
• Continuous planning in Linear, 5 mins
• Magical Thinking on AI, 10 mins