Beauty, Breadth, and Volume

1. Energy Costs

Decarbonisation and sustainability have been key factors driving design. Green building councils, legal mandates, and the professional care that designers have for the environment have driven an increase in sustainable thinking.

The solar revolution is changing the way I think about this. Adam Tooze pointed out this week that the capital cost of solar is now competitive with fossil fuel power generation. And the opex is close to zero thereafter. Power is getting cheaper and greener. Insanely cheap.

What this means is that the operational carbon intensity of a building is falling massively.

The embodied energy, or the upfront carbon as I've seen it called, needs to be accounted for.

But the reality right now is that you can build a leaky, uninsulated house and run the air conditioning for free, and still export power with a solar system that costs ~$4K. The economics for commercial buildings are not massively different.

The change is that solar panels are cheaper than roofing and are being used as fencing.

The input assumptions of decarbonisation business cases are swinging wildly.

Something that will never change, though, is that buildings that are beautiful will be kept. And things that are kept for many years have a greater denominator in per-year emissions.

Building cities that are lovely seems the most reliable predictor of sustainability — and that goal will not change with the energy economics.

2. Marketing a Broad Tool

Giraffe has always had a marketing problem. The product's power and flexibility is a marketing headache. We put out videos of capability and features, but they land like snowflakes on a roof. Individually great, but completely disconnected.

The magic of Giraffe is that it can do anything. It has a focus: cities. But in that broad sector it is extremely wide-ranging, doing anything from policy design to facade design.

This breadth has often been seen as a weakness — it doesn't lend itself to an elevator pitch. It can be perceived as a lack of focus.

But what we've been trying to build is a tool that can orchestrate workflows across the stack, because the sector is broad and in desperate need of new solutions. It has seen zero productivity gains in decades.

More than this, it is the critical sector. Humanity, increasingly, is an urban species: and all the activities of our species — work, transport, play, housing — happen within the urban environment.

One way of trying to visualise Giraffe's complexity is to forget the product and just map out the workflows. In this broad sector, the workflows are very broad — across disciplines, firms, and even paradigms. But once you've started mapping the workflow, it is easier to see where software of any kind, including Giraffe, fits together.

To make that concrete, I've built an app at giraffe.build/orchestrator.

This video introduces it:

The iframe below shows one workflow as it is implemented concretely in Giraffe — concept plan, landscape plan, financial analysis, facade design, and leasing and marketing material:

3. LLM Productivity

I've made 27 apps since the start of March. Roughly one a day. This has not majorly disturbed my primary role — which is heavily product and R&D focused — but it is extraordinary.

This week it has been brought home to us a few times that LLMs don't always provide excellent answers.

Watch Andrew Wallace's (TCC) nuanced, experienced explanation of his use of AI here.

LLMs do, however, write great code, and code can provide good answers.

Why? Code is "self-healing" — it throws errors if it is wrong, which eliminates one class of bugs. Obviously major bugs exist, but at least the code is transparent: you can see and inspect it.

The LLM will separate business logic from UI and state management and provide very clear comment into what it is doing, if you ask it to.

Also, most code is doing very simple things — it is just doing lots of them.