Menu
Close
.svg)
As AI becomes embedded across the clinical workflow, pharma marketers need to rethink how information and evidence reaches healthcare professionals.
Earlier this month, Eli Lilly announced a strategic investment in Abridge, one of the leading companies in ambient clinical documentation. It would be easy to file this under “another AI healthcare deal” and move on. That would miss the more interesting point.
Pharma has perpetually invested in the machinery of influence and evidence, through R&D, publications, congresses, medical education, patient support and KOL engagement. The assumption has been that physicians encounter evidence through fairly recognizable channels — whether a journal article, a society meeting, a guideline update, a colleague, a rep or MSL.
But that assumption is starting to look incomplete.
Ambient documentation and dictation is becoming one component of a broader AI layer forming around clinical care. It’s important for pharmaceutical companies to pay attention to it because this layer increasingly sits between the medical information that physicians receive and the decisions they make.
Unlike many earlier healthcare technology trends, AI is already finding its way into physicians’ working days, whether it’s administrative or clinical. Physicians have too much information to process and too little time in which to process it. This is not a new problem, of course; it’s just one for which AI technology providers have arrived with unusually good timing.
I run a healthcare marketing research agency but still spend 20 to 30% of my time speaking directly with HCPs as part of primary marketing research studies. One question I now regularly pose up front during interviews is about their “information diet,” i.e., the mix of journals, websites, conferences, colleagues and tools HCPs rely on to stay current and make clinical decisions.
For years, the answers had a dependable pattern. Physicians cited specialty journals, society meetings, Medscape, UpToDate, PubMed, trusted colleagues and reps. Recently, an academic cardiologist gave me a different answer: She added OpenEvidence.
When I asked why, she didn’t give a grand theory of medical AI. She said: “It’s conversational, and now that it’s pulling from sources like NEJM and JAMA, it helps me get to the evidence faster.”
That is the market shift in miniature. Other platforms will appear and gain share and influence. Some will be acquired, rebranded or forgotten, as is customary; but the point is that physicians are beginning to treat AI tools as part of the route by which evidence reaches them.
The American Medical Association (AMA)’s 2026 Augmented Intelligence physician survey suggests this is not an isolated habit. Nearly 40% of physicians now report using AI tools to summarize medical research and standards of care within their workflows, making it one of the most common clinical AI use cases today. More than 80% report active usage of AI in practice.
The usual way to discuss healthcare AI tools is to divide it into categories: ambient scribes, clinical decision support, evidence synthesis, EHR copilots and prior-authorization tools. Those categories are useful for investors, vendors and conference panels. They are less useful for understanding how physicians actually experience their work.
Physicians are not moving through their day with the “ambient documentation journey” in mind. They are simply attempting to get through the clinic, stay current, make sound decisions, document appropriately, avoid missing something important and go home at a reasonable hour. AI is beginning to attach itself to each of those tasks.
Before an encounter, AI may help physicians prepare by summarizing relevant medical history, highlighting care gaps, and prioritizing patients who may require additional attention or intervention. During the visit, ambient technologies will automatically document the conversation while decision-support tools provide reminder prompts or recommendations. Afterward, AI may assist with coding, prior auth, follow-up communications and, increasingly, identification of appropriate clinical trials.
None of this removes the physician from the center of care. It does, however, alter the environment around the physician. The exam room is no longer occupied only by doctor, patient and EHR. Increasingly, there is an additional participant — not making the decision per se, but shaping what is seen, summarized and remembered.
For decades, commercial and medical teams have asked familiar questions: Where should our data be published? Which congresses matter? Which journals are most influential?
These questions remain important and these sources aren’t being replaced, but they now sit next to a newer question: How does our evidence appear inside the AI-enabled tools physicians increasingly utilize to navigate medicine?
When physicians searched PubMed, they interacted directly with the literature, at least in principle. They still had to arbitrate quality, relevance and applicability. But the encounter was with the paper itself. Increasingly, physicians are interacting with systems that summarize, rank, contextualize and compare evidence before the physician sees it in full.
The evidence may be published in the right journal, presented at the right congress and discussed by the right experts. Yet its practical influence may also depend on whether it’s retrievable, interpretable and correctly represented inside systems that were not built by pharma and are not controlled by pharma.
Patients are entering this picture too. A recent STAT article described patient-facing tools that record visits, generate summaries and help patients navigate care plans after appointments.
It’s not hard to imagine a near future in which both sides of the encounter are supported by AI systems. The physician’s tool summarizes the visit one way, and the patient’s tool summarizes it another. Both could be useful, but also incomplete.
For pharmaceutical companies, here’s the implication: The industry has spent decades studying how physicians make decisions. It now needs to study how physicians make decisions with AI in the exam room.
That doesn’t mean chasing every new platform or trying to “optimize for AI” in some vague search-engine sense. It means understanding how scientific information is ingested, summarized and surfaced by the tools physicians are beginning to use. It means asking whether pivotal evidence is represented accurately. It means learning where AI changes the path from publication to perception to practice.
Physician information diets are changing, and pharma doesn’t need to panic. It does, however, need to move beyond the assumption that traditional channels alone determine how evidence is discovered, interpreted and applied in practice.
