Working inside the idea

At a glance TL;DR

• The CAD/CAM habit of thinking through a model could become accessible far beyond specialist design work.
• People and agents could develop proposals together in a shared environment connected to the tools for realising them.
• Lower production and coordination costs could change how work is organised and give smaller organisations access to much deeper investigation.

A great deal of working life is spent describing things that do not yet exist. A new building. A reorganised factory. A better way to run a café. We turn intentions into briefs, diagrams, spreadsheets and presentations, then ask other people to reconstruct what we mean.

AI points towards a different arrangement. The proposal itself becomes the place where people and agents work together: developing an idea, exploring its consequences and directing the work needed to realise it.

In a recent Moonshots conversation, Dave Blundin used CAD/CAM—computer-aided design and manufacturing—to explain the possibility. For people who use these systems, the model is part of how they think. They develop an idea by working on it, examining it and changing it. Blundin’s suggestion was that this experience could spread far beyond specialist design and engineering.

The immediate prompt was Google’s Genie 4 announcement. Its demonstrations show generated environments that respond to movement and language, changing as the user interacts with them. The research preview makes a new kind of interface tangible: a scene you can keep shaping while you explore it.

A convincing scene is not yet a dependable forecast. Google acknowledges that details can drift, real locations are imperfectly represented and continuous interactions last only a few minutes. The pace of improvement at the frontier suggests that useful versions of a connected working environment may be close. To explore that future, assume the interface is connected to dependable models, relevant information and tools that can carry decisions into action.

Take a team planning a café. The owner wants a welcoming place where people linger, while keeping the morning takeaway queue moving. Today, that intention travels through conversations, layouts, equipment lists and financial assumptions. Each describes part of the same business.

In this future, the owner, designer and café manager work through a busy morning together in a shared model. Their agents develop layouts, investigate equipment choices and test different service arrangements. The manager points out where used cups will accumulate. The designer explores a different collection point. The owner asks what happens if the queue doubles. Agents work through the consequences and bring alternatives into the conversation.

The original question turns out to be incomplete. The counter position matters, but so does the sequence of ordering, waiting and collecting. A promising solution combines a spatial change with a different way of serving. Design and operation develop together because people can examine their relationship while both remain changeable.

Between conversations, the agents continue investigating. They compare supplier information, resolve dependencies and return with discoveries: a cheaper equipment arrangement, an overlooked access problem, a way to preserve the quiet seating area. The work advances between human contributions, with decisions and unresolved questions brought back into view.

More knowledge can enter the process. Someone who has worked behind a counter for ten years may recognise a problem immediately but struggle to communicate it through a plan. A responsive model gives that experience somewhere to act. Participation depends less on knowing how to operate the software or commission another presentation.

The brief develops through the work. Seeing a crowded entrance reveals that the owner’s idea of hospitality includes room to pause, even at the expense of another table. People discover what they want through encounters with possibilities. The agents contribute ideas too, helping the team reach outcomes nobody had fully specified at the outset.

The connection to making is fundamental to the CAD/CAM analogy. The shared model becomes a means of directing the work. An agreed change carries through into equipment requirements, cost comparisons, drawings, orders and work instructions. Agents resolve dependencies and return consequential choices to the people involved. The conversation leaves behind both a clearer intention and progress towards realising it.

The organisational implications are substantial. Much coordination exists because information and intentions are distributed across separate systems. Connecting them removes work previously needed to translate, report and reconcile the same decision. A practice or business can reconsider how it divides responsibilities when those burdens diminish. Expertise remains active throughout, but the work surrounding it changes.

The economics change who can attempt what. Our café owner gains access to a depth of design and operational investigation previously difficult to afford. A small practice can explore more alternatives without staffing a large production team. More ideas become worth pursuing when the cost of developing them falls. Rent, materials and construction still have to be paid for; the substantial change is in the intelligence and effort available to shape what gets made.

This is a larger vision of work: people and agents forming intentions together, developing them through a shared representation, and carrying them into the world. The distance between having an idea and being able to do something serious with it becomes much smaller.

Sources

1. Moonshots episode 307: AI's Stock Market Fallout, Google's Profit Problem, and Altman's Hold My Beer Moment Retrieved 2 November 2026
2. Google DeepMind: Genie 4, a new generation of world simulation, 27 October 2026 Retrieved 2 November 2026
3. Google DeepMind: Genie 4 capabilities and limitations Retrieved 2 November 2026