Officials have discussed paying tech-company AI 60% to 80% of what a physician earns for the same service — and 200+ companies are already inside Medicare pilots that can involve AI

Why it matters for an independent practice: This is the most direct threat to practice economics the feed has carried this year, and it is not a capability story — it's a price story. If a payer will reimburse software at 60–80% of the physician rate for the same service, the ceiling on what that service is worth has been set by someone who has never met your patient, and every downstream negotiation starts from there. Two concrete things for an independent practice or an MMR client: one, the FDA pathway that lets four unauthorized models reach Medicare patients is the pathway your competitors' vendors are lining up for, so "is it FDA-cleared?" is about to have a third answer that isn't yes or no — it's "it's in a pilot." Two, this is where yesterday's procurement framework earns its keep: question 1 was FDA clearance and the precise indications for use, word for word. A pilot admission is not a clearance, and a sales deck will not draw that distinction for you.

Why it matters for an independent practice: Strip out the enterprise framing and this is a claim about adherence, which is the single most under-measured variable in regenerative medicine. PRP, BMAC and MFAT outcomes turn on what the patient does in the weeks after the injection, and the patients who don't do it are disproportionately the ones carrying something nobody screened for. The buildable version for an independent practice is smaller than the health-plan version and costs almost nothing: ask the adherence question at the follow-up, not just the pain-score question, and keep a live list of who you can actually refer to with current wait times. The closed-loop point is the part worth stealing outright — most practices have no idea whether a referral they made ever happened, and item 1 says you will eventually be asked to produce outcome data to defend your prices.

Why it matters for an independent practice: This is the hardest evidence yet that the AEO work MMR sells cannot be done on the client's own website. If 85.8% of what gets a brand quoted comes from third-party pages, then the deliverable isn't another service page — it's getting the practice onto the "best regenerative medicine clinics in [city]" lists that the models are already pulling from, and keeping it there, because half of those placements decay inside a month. Three things that follow directly: one, run the prospecting step — check which sources the models actually cite for your client's money keywords before commissioning anything, because placement on an already-cited source is what compounded. Two, budget for maintenance, not launch; a one-time placement is a depreciating asset. Three, the heading finding is free to implement this week — rewrite comparison headings as the question a patient would actually ask.

Why it matters for an independent practice: Zero medicine in this, and it is directly usable in two ways. First, the procurement angle, which is the whole point: yesterday's item was a physician's framework built on getting a vendor to commit in writing and noting carefully what they decline to sign. Here is the single largest vendor in a medical practice's stack — Microsoft 365, Teams, the identity every one of your systems hangs off — publishing exactly that document, unprompted. You can now ask any smaller AI vendor the same question and point at the example. The ones who have nothing comparable will say so by changing the subject. Second, the deepfake line is not abstract for this audience: physician impersonation is an active problem — Medical Economics filed a video on it on Sep 2 — and a top-tier vendor declaring synthetic-likeness generation an absolute constraint that a paying user cannot override is a concrete trust asset. The honest caution: this is a self-published, self-enforced document with no external auditor named, which is precisely why "embedded evaluators" is the phrase to watch. A rule you write about yourself is a marketing claim until someone else can check it.

Why it matters for an independent practice: The pitch every practice is about to receive is "an AI agent for your front desk." The most interesting thing an Anthropic engineer can say into that market is don't build the agent — and the reason is almost certainly the one that matters for a clinic: an agent is a system that decides what to do next, and in a practice the expensive failures are the ones nobody specified. A deterministic workflow that always does the same four things is cheaper, auditable, and explains itself to a patient who asks what happened. Watch this before the next vendor demo, and hold it against yesterday's question 5 — what happens on failure, and how is historical output reviewed? An agent makes that question much harder to answer, which is a reason to buy one deliberately rather than by default.