The operating philosophy
Human in Presence™
The human is present, not in the loop. Machine-speed execution, with human judgment held exactly where it belongs.
The problem with the loop
The dominant pattern for keeping people in control of AI is Human in the Loop (HITL): a person sits in the path of every action and approves it before it proceeds. It sounds responsible. In practice it fails twice.
It does not scale — a person who must approve every routine step becomes the bottleneck the whole system waits on. And it misplaces the one thing it was meant to protect: human judgment is spent rubber-stamping the trivial, and is exhausted by the time something actually requires it.
Two ways to keep the human in control
Human in the Loop
A gatekeeper on every action
Authority lives in the person. Nothing proceeds without an approval. Judgment is spent on volume; accountability blurs into a stream of clicks; the system runs at the speed of the most overloaded human in it.
Human in Presence
An accountable witness over the whole
Authority lives in the system, established up front. Agents act within it at machine speed. The human is present as witness, exception-handler, escalation target, and the one accountable — spending judgment only where judgment is required.
Why this is the right philosophy for care
Care is intimate, regulated, and high-stakes. Accountability cannot be automated away — someone must be answerable for what happens to a person. But neither can a human meaningfully approve every medication reminder, every schedule change, every routine claim. Asked to do so, they stop reading, and the approval becomes theatre.
Human in Presence holds both truths at once. The work runs at the speed the system can deliver it. The human remains genuinely accountable — present for the exception that matters, not buried under the thousand that don't.
The human is present, not in the loop.