Delphic Biosystems had been too generous by far. Not only had they arranged for me to interview ten times as many of their Public Relations staff as I could ever have made time for, they’d showered me with ROMs packed with seductive micrographs and dazzling animation. Software flow-charts for the HealthGuard implant were rendered as airbrushed fantasies of impossible chromed machines, jet-black conveyor belts moving incandescent silver nuggets of “data” from subprocess to subprocess. Molecular schematics of interacting proteins were shrouded with delicately beautiful – and utterly gratuitous – electron-density maps, veils of pink and blue aurorae melting and merging, transforming the humblest chemical wedding into a microcosmic fantasia. I could have set it all to Wagner – or Blake – and flogged it to members of Mystical Renaissance, to play on a loop whenever they wanted to go slack-jawed with numinous incomprehension.

I slogged my way through the whole morass, though – and it finally paid off. Buried amongst all the technoporn and science-as-psychedelia were a few shots worth salvaging.

The HealthGuard implant employed the latest programmable assay chip: an array of elaborate proteins bound to silicon, in many respects like a pharm’s synthesizer, but designed to count molecules, not make them. The previous generation of chips had used a multitude of highly specific antibodies, Y-shaped proteins planted in the semiconductor in a chequerboard pattern, like adjoining fields of a hundred different crops. When a molecule of cholesterol, or insulin, or whatever, happened to strike exactly the right field and collide with a matching antibody, it bound to it long enough for the tiny change in capacitance to be detected, and logged in a microprocessor. Over time, this record of serendipitous collisions yielded the amount of each substance in the blood.

The new sensors used a protein which was more like a Venus flytrap with brains than an antibody’s passive, single-purpose template. “Assayin” in its receptive state was a long, bell-shaped molecule, a tube opening out into a broad funnel. This conformation was metastable; the charge distribution on the molecule rendered it exquisitely sensitive, spring-loaded. Anything large enough colliding with the inner surface of the funnel caused a lightning-fast wave of deformation, engulfing and shrink-wrapping the intruder. The microprocessor, noting the sprung trap, could then probe the captive molecule by searching for a shape of the assayin which imprisoned it even more snugly. There were no more wasted, mismatched collisions – no more insulin molecules striking cholesterol antibodies, yielding no information at all. Assayin always knew what had hit it.