Where AI gets human.

284 thoughtful people live now
Y
Fresh voicesCreators building practical, accountable AI.
vlogAI in practice

I gave 12 nurses an AI scribe. Here’s what we learned.

The biggest win wasn’t faster notes. It was getting eye contact back. But the consent flow needed a complete rethink.

blogGovernance

The case for boring AI governance

Good governance is rarely cinematic. It looks like clear escalation paths, versioned decisions, procurement checklists, and somebody empowered to say no. I mapped the operating layer every organization can start with.

OPERATING NOTE 04
“Accountability is a workflow, not a value statement.”
CS
chatPrompt clinic

Can an AI explain a benefits denial without giving legal advice?

A redacted, annotated chat showing how our team narrowed the task, added source boundaries, and designed a safe handoff to a human caseworker.

● ● ●REDACTED CHAT · 14 TURNS
CASEWORKERExplain the reason in plain language, citing only the supplied letter.
ASSISTANTI can summarize the stated reason and flag what needs a qualified reviewer…
podcastNew episode

Who gets to decide what an AI refuses?

A conversation with a civil liberties lawyer and a safety researcher about model boundaries, appeals, and accountability.

Podcast microphone in a studio
00:00 / 42:1814.2K listens