Uber says only four rides were booked by an AI agent during its busiest week of October
November 5, 2026 · 3 items
Uber says only four rides were booked by an AI agent during its busiest week of October
TechCrunch · November 4, 2026Buildable AI
On Uber's third-quarter earnings call Wednesday, the company said just four rides were ordered through an AI agent during its busiest week in October — the week of Halloween.
Agent-driven ordering was somewhat more common on the food side: TechCrunch reports more people used an AI agent to order through Uber Eats than to book a ride.
Reported by Sarah Perez, published 1:56 PM PST on November 4, 2026.
Why it matters for an independent practice: This is the cleanest public number yet on real consumer agent adoption, and it should set the pricing of any 'agentic booking' pitch a practice hears this quarter. If one of the largest consumer transaction apps in the country counts agent-initiated bookings in single digits at peak, patients are still calling, texting and clicking — so fund the human-facing intake and scheduling path first, and treat agent integrations as a cheap experiment to be ready for, not a channel to migrate to.
Shuang-Ho Hospital deputy director says a hospital-built risk model lets teams catch acute kidney injury early
Healthcare IT News · November 4, 2026Practice operations
A HIMSS TV clip posted November 4, 2026 at 4:33 PM features Shuang-Ho Hospital deputy director Dr Meng-Che Lu on an AI the Taipei hospital built to flag kidney injury risk.
The page's description says the tool 'has allowed clinical teams to spot acute kidney injury early'; it's filed under HIMSS26 APAC and tagged Clinical, Decision Support, and Quality and Safety.
Thinness disclosed: this is the publisher's own description, not a transcript — no runtime, model details, validation method or outcome numbers are stated on the page.
Why it matters for an independent practice: This is the shape of clinical AI worth pointing a physician client at when they ask what exists beyond ambient scribes: a narrow, condition-specific risk model built and owned inside the institution. The transferable move for an independent practice evaluating any diagnostic or risk-flagging tool is to ask the same question this clip raises and doesn't answer — on whose patients was it validated, and what happens to the workflow when it fires.
Only 22 of 166 AI-and-health-equity papers actually defined health equity, a Georgetown scoping review found
Medical Economics · November 3, 2026Practice operations
Georgetown University researchers searched four databases and found that just 22 of 166 peer-reviewed articles on AI and health equity included an explicit definition of health equity.
Medical Economics reports the lead author said that gap could steer resources toward the most advantaged groups; most guidance was also characterized as general and U.S.-based.
Written by Austin Littrell, fact checked by Keith A. Reynolds, bylined November 3, 2026.
Recency flag, stated plainly: the underlying scoping review was published October 8 in Health Affairs Scholar — older than this feed's normal window, and run here because the finding is a durable question rather than a news event.
Why it matters for an independent practice: It converts into one sentence you can put in front of any AI vendor courting a practice: 'Which definition of equitable access are you measuring against, and on what population?' When a scheduling, triage or outreach tool claims it widens access, a vendor with no answer is selling a claim that most of the literature apparently also failed to pin down — and the practice carries the outcome, not the vendor.
