Guillaume Ardaud spent eight years at Apple prototyping Face ID, Apple Pencil, and the iPhone camera system. He knows how devices that live on your body get built — and apparently he'd rather not build another one. His Lightbulb Computer project, published September 2026, is a speculative design exercise that asks a simple question: what if spatial computing projected outward into the room instead of inward onto your retinas? The concept is a large-bulb-shaped device housing a projector and computer vision system, designed to screw into any standard Edison socket or sit in a portable battery base. It sees the room, listens for voice commands, tracks where you point, and throws interactive content onto whatever surface is available — kitchen counters, desk surfaces, walls, even a 3D terrain map. The demos shown are real prototypes running on bulky hardware, not renders or AI-generated video. That honesty is the project's strongest quality. The use cases Ardaud walks through are deliberately domestic and low-stakes: pinning a recipe to a countertop while your hands are covered in flour, projecting a shared map for trip planning instead of huddling over a phone, ambient bedside displays showing weather and alarms, pointing at a smart home device to control it instead of remembering its name. None of these are technically novel — projector-based interaction research goes back decades (I/O Bulb, LuminAR, Dynamicland, Folk Computer). What Ardaud is doing is reframing that research tradition as a personal computing proposition for homes and offices, not labs and museums. The critique of eyewear-based computing is the sharpest section. Ardaud points out the compounding constraints: interference with hair, makeup, jewelry, existing glasses; severe power and thermal limits from miniaturization; the fundamental antisocial quality of requiring every person in a room to wear compatible hardware. He includes a devastating pair of images — the marketing vision of AR glasses showing digital overlays in a living room, versus what you'd actually see walking into that room: people wearing goggles staring at nothing. Where the project is weakest is where Ardaud himself flags it: the gap between prototype and product. Projector brightness in ambient light, resolution at throw distance, occlusion handling, latency for gesture tracking, and privacy architecture are all unsolved at consumer scale. The privacy section is a single paragraph of hand-waving about 'existing best practices' and hardware indicators — the thinnest part of an otherwise thoughtful piece. A ceiling-mounted camera with always-available activation in every room of your house is a harder privacy problem than a phone in your pocket, and deserving more than a sentence. The piece works best as design criticism rather than product pitch. Ardaud is not claiming to have built a shipping product; he's arguing that the industry's entire vector — face-mounted displays — is a wrong turn, and that a different vector exists with real advantages for social, ambient, and shared computing. That argument is well-made and historically grounded, even if the engineering path forward remains steep.
