What if the problem with street lighting is not the fixture, but the whole way we design it?

In this conversation on Light Talk, Martin Klaasen sits down with Croatian lighting designer and developer Dean Skira, the creator of Taman. Dean traces his path from early work in New York to public projects in Pula, including the city’s celebrated lit cranes and its first light festival.

Dean explains how Taman flips the workflow from product-first to requirement-first. It starts from a simple Google Earth snapshot and hand-drawn layers, calculates the whole urban environment rather than just the space between two poles, positions the fittings, optimises photometric curves and compares market-ready products with the ideal result. He also talks about the role of machine learning in the software, and a real project where the luminous flux came down by 45.8%.

The conversation then turns to the bigger picture: outdated manufacturing, political decision-making, copycat products, master planning for whole districts and cities, and why the future may belong to modular, mass-customised luminaires.

Essential listening if you work in lighting design, urban planning, street lighting or product development.

In this Light Talk bonus, Daniel Green gives Martin Klaasen a hands-on tour of Dusk, his lighting-specific AI agent that turns rough sketches, reference photos and text prompts into lighting concepts, and of Lumviewer, his photometric tool for comparing luminaires from LDT and IES files. Daniel explains why he built Dusk as a dedicated lighting agent rather than relying on generic tools, shows how it handles beam angles, colour temperature, wall grazing, focal glow and realistic placement, and introduces Dusk GPT and Dusk Studio, the local version in development. It follows the full episode, Daniel Green: Transforming Lighting Design with AI. Links: dusk-ai.co.uk and lumviewer.replit.app