VA opens a $775.72M five-year ambient scribe contract to all its medical centers, with Abridge already live at 75+ sites
September 24, 2026 · 4 items
VA opens a $775.72M five-year ambient scribe contract to all its medical centers, with Abridge already live at 75+ sites
Healthcare IT News · September 23, 2026Practice operations
The VA selected Abridge AI to compete as an ambient AI vendor available to every VA medical center under an enterprise contract; the multiple-award ambient scribe AI contract carries a total ceiling of $775.72 million over five years across all eligible vendors.
Abridge says it is fully operational on both VistA/CPRS and the new Oracle Federal EHR, and in clinical use at more than 75 VA medical centers.
In April the company added peer-reviewed research from NEJM and the JAMA Network, plus Wolters Kluwer's UpToDate, into its clinical decision support so documentation cites evidence-based sources during the visit.
Prior steps: a 10-site pilot expanded to 130+ medical centers via Rise8 and Thoughtworks Federal in March, and as of June the VHA had deployed ambient scribe to all PACT primary care providers.
Why it matters for an independent practice: The largest integrated health system in the country has now made ambient documentation a procurement line item, not an experiment — that normalizes the tech for every independent practice patient and staff member, and it sets the bar your scribe vendor should be measured against: works inside your actual EHR, cites real sources, and survives an EHR migration.
ARPA-H commits up to $62.7M to build patient-facing autonomous cardiac AI agents — and a separate AI to supervise them
Healthcare IT News · September 21, 2026Practice operations
The ADVOCATE program commits up to $33.7 million in year one of a four-year, $62.7 million effort; Atman Health, Tempus AI and UpDoc build the patient-facing agents, and the cohort must submit a first-of-its-kind FDA authorization package within 24 months of award.
Stanford is building a disease-agnostic supervisory AI agent that continuously monitors the clinical agents post-deployment for unsafe recommendations and out-of-distribution behavior, via a three-stage pipeline producing "inspectable per-claim rationales."
Duke will run multi-site validation across five health systems and rural sites on both Epic and Oracle EHRs; Kaiser Permanente will embed the agents across 21 medical centers and more than 260 clinics; Johns Hopkins APL evaluates independently.
ARPA-H cites more than 200,000 preventable cardiovascular deaths a year and notes nearly half of U.S. counties lack cardiology providers; FDA Digital Health Center of Excellence director Dr. Rick Abramson called patient-facing agentic AI "one of the most consequential frontiers in digital health."
Why it matters for an independent practice: This is the regulatory template for autonomous patient-facing AI being written in public — when the FDA framework lands, it will define what an independent clinic is allowed to let an agent say to a patient between visits, and the "supervisory agent that audits the agent" architecture is the right mental model for any practice deploying intake or follow-up automation today.
Healthcare Success: AI systems now interpret your site as a dataset, and technical SEO is the visibility infrastructure that decides whether you're surfaced at all
Healthcare Success · September 23, 2026Patient acquisition
Author Brandon Schakola argues search engines have shifted from directories that point users to pages toward interpreters that extract meaning and present synthesized answers — so a site must be "machine-readable, semantically clear, and internally consistent" to be included in an AI Overview at all.
He cites Google Search Central guidance that modern search systems use structured data, page hierarchy, and contextual signals to determine relevance and reliability, not just indexed URLs.
The framing that matters for clients: "Your brand isn't just a website, it's a dataset — some of which you control, and some of which is floating out off-site," and most brands have never audited how fragmented their identity has become.
Healthcare is classified as YMYL content, which raises the bar the AI systems apply when comparing your signals against known healthcare entities.
Why it matters for an independent practice: This is the concrete MMR work order: schema, entity consistency, and off-site identity cleanup are no longer hygiene tasks but the precondition for a regen-med practice showing up in an AI answer — audit each client's structured data and off-site listings against the physician/practice entity before spending another dollar on content.
ChatGPT's voice agent lands on mobile: draft docs, summarize Slack, resume on desktop — for Plus and Pro
TechCrunch · September 23, 2026Buildable AI
OpenAI announced Wednesday that Plus and Pro users can use the Work tab on their phone to create a document, draft an email, or summarize Slack messages by voice; they can also build sites, create presentations, and use the cloud browser.
Free and Go users get plugins and connected apps rather than the full agentic Work tab.
Voice conversations now produce richer text output, users can switch between text and voice, and a conversation started on the go can be resumed on desktop.
This extends GPT-Live, the conversational model launched in July and integrated into the desktop app later that month; Anthropic recently merged its Cowork and Chat interfaces, while OpenAI keeps chat and workspaces separate.
Why it matters for an independent practice: The buildable application: a physician-owner can dictate the raw substance of a blog post, a referral letter, or a patient-education explainer between cases and have a draft waiting at the desktop — but the stewardship question comes first, because the same Work tab reaches connected apps, so decide what a voice agent is allowed to touch before anyone in the practice points it at anything patient-adjacent.
