Software is getting cheaper, orchestration is still hard.

Software is Getting Cheaper. Orchestration is Still Hard.

A few months ago I needed a graphic of a city for the Giraffe website. Simple enough. I opened Figma — the tool every designer reaches for — and started drawing.

Then I realised I needed hundreds of buildings.

Figma is excellent at drawing one rectangle. It is not excellent at drawing four hundred. Four hundred rectangles, in short, was not going to happen. But I had a thought that $500b in AI investment might mean that I could use AI to build a tool that automated the rectangles faster than I could draw them manually.

Build the Tool You Need

So I closed Figma, and opened Claude. It built a small web applet that made cities. You can see it below. It created a library of building typologies rendered in axonometric view — each building made of four or five SVG shapes. It let me click on a square, and place a building, and when I asked even automatically populated it with buildings, then trees. It even did roads, and animation! What would have taken days of manual drawing took minutes.

The tool I needed didn't exist, so I built it.

AI has changed the cost equation. The barrier to fixing a broken workflow isn't cost — it's perception.

The economics of software are changing. For twenty years the default answer to a workflow problem was to buy another SaaS product. Increasingly the answer will be to generate a small piece of software that solves the exact problem — and run it inside a platform that connects it to real data, permissions, and governance.

This shift is already being debated across the software industry. Silicon Valley has started referring to the phenomenon as the “SaaSpocalypse” — the idea that AI agents may compress the value of traditional SaaS applications.

Developer communities are asking similar questions. Discussions on Y Combinator’s Hacker News frequently explore whether companies will increasingly generate internal tools rather than relying entirely on packaged software.

A Fragmented Industry

Property development is a sector in need of retooling. It is one of the least digitised industries in the white collar world, not because developers are unsophisticated, but because the industry is extraordinarily fragmented.

Every project is different. Every council is different. Data lives in dozens of disconnected places. There is no single system of record.

So developers do what people always do when systems fail them: they improvise with whatever's nearby.

That usually means Excel and Google Maps.

The Manual Workflow

Assessing one development site in Excel is manageable. Assessing ten is a slow grind. You often don't know which of those ten sites is actually profitable until you've spent days finding out.

For each one, you're manually pulling:

• Parcel data
• Zoning
• Slope
• Heritage overlays
• Flood risk
• Height of building controls
• Setbacks

Seven data layers. Ten sites. That's seventy manual lookups, cross-referenced and re-entered by hand, with every attendant risk of error and every hour spent doing it instead of thinking.

This isn't a modelling problem. It's a data orchestration problem.

The Platform Approach

The information exists. The analysis is well understood. What's missing is the connective tissue — a way to pull the right data, in the right shape, for the specific question you're trying to answer right now.

Giraffe is built for exactly this. Instead of buying a monolithic platform that half-fits your workflow, you build micro-apps: small, precise tools that sit inside your existing data ecosystem and do one thing well.

• A site sieve that runs a candidate address against twenty or thirty data layers in minutes rather than days
• A communications tool that generates marketing materials from project data
• A feasibility model that updates in real time as inputs change

The key word is build. Not buy.

AI Changes the Build Equation

AI changes the build versus grind-it-out equation — and in fragmented industries this shift is particularly dramatic. When software becomes cheap to generate, the scarce capability becomes orchestration: defining the process, structuring the data, and governing how systems interact.

AI can now write these micro-apps. A developer with a clear understanding of their problem — not a vague sense of friction, but a precise diagnosis — can describe what they need and have a working tool in hours.

The barrier to custom software has collapsed.

Why the Platform Still Matters

An AI-generated app running in isolation is fragile. It has no access to your real data. It has no guardrails. It has no understanding of the constraints your business actually operates under.

Giraffe provides not just the ability to build micro-apps, but the data infrastructure and boundaries that make those apps trustworthy.

AI provides the leverage. The platform provides the foundation.

A property developer can build precisely the tool they need — anchored to real data — without a six-figure SaaS contract and an eighteen-month implementation.

The First Step

Some of the most expensive friction in your business right now is a process problem or a data orchestration problem — unaddressed because no one has realised how quickly it could be rebuilt.

The first step is believing it's solvable.

Click the Grow button to bring the city to life.