No PriorsNo Priors Ep. 130 | With OpenEvidence Founder Daniel Nadler
CHAPTERS
- 0:05 – 1:54
OpenEvidence’s rapid adoption as clinical decision support
Daniel Nadler frames OpenEvidence’s growth in the context of compressed AI adoption cycles and claims it has become a default tool for high-stakes clinical decision support. He distinguishes this category from lower-stakes workflow tools like scribing or paperwork, emphasizing the one-shot nature of patient care decisions.
- •AI adoption cycles are compressing across knowledge work, including medicine
- •OpenEvidence positioned as an 'operating system' for clinical knowledge
- •High-stakes clinical decision support differs from administrative workflow tools
- •Errors in clinical decisions can’t be easily undone compared to documentation tasks
- 1:54 – 4:04
How it works: semantic search for complex patient scenarios
Nadler explains OpenEvidence as a semantic search engine rather than an 'answer engine.' He walks through a detailed psoriasis-plus-MS example to show how clinical queries are paragraph-length and require deep contextual understanding to retrieve the right evidence snippets.
- •Clinical decision support is fundamentally a semantic search problem
- •Medical queries often require multiple sentences of context
- •Example: choosing IL-17 vs IL-23 inhibitors when a patient has MS
- •Goal is to map scenario semantics to precise evidence in biomedical literature
- 4:04 – 6:35
Reducing medical error by routing to primary evidence
The conversation connects search quality to real-world harm reduction, arguing that misinformed treatment choices can worsen comorbidities even when they don’t cause death. Nadler describes the product’s core as locating canonical trials/guidelines and the specific passages that answer the clinician’s question.
- •Medical error impacts morbidity far beyond mortality statistics
- •Correct answers may live in non-obvious sections (methods/population)
- •System targets landmark RCTs and guidelines from a massive corpus
- •Value comes from fast, precise retrieval for time-constrained clinicians
- 6:35 – 9:16
Handling ambiguity: physicians as users, transparency, and citations
Elad asks how OpenEvidence deals with conflicting or immature evidence. Nadler argues that limiting the product to physician users enables a safer 'tooling' model, and that the UI should explicitly surface uncertainty and conflicting sources while making citations central to trust and auditability.
- •Strategic choice: serve physicians first, not patients
- •Surface ambiguity and contradictory evidence rather than asserting certainty
- •Citations and references are a first-class feature for auditability
- •Framing as a search tool sets expectations and the user 'social contract'
- 9:16 – 11:37
How doctors use it in practice: audit-first behavior and referral to journals
Sarah probes whether users actually read sources; Nadler says they do, routinely. He describes workflows where clinicians ask complex questions, receive a shortlist of canonical papers, and click through—making OpenEvidence a major referrer to journals like NEJM.
- •Typical flow: complex query → 3–5 key sources → click through to originals
- •Emphasis on routing to evidence, not providing final answers
- •Large corpus (tens of millions of publications) narrowed to actionable snippets
- •Partnerships with top medical journals strengthen trust and data quality
- 11:37 – 15:53
Treating physicians like consumers: the app-store distribution hack
Sarah asks why OpenEvidence chose physicians vs consumers; Nadler explains he wanted a consumer-internet growth model applied to knowledge workers. He argues doctors had been treated as extensions of health systems, and that simply letting them download a free app on their own phones unlocked adoption.
- •Goal: build a consumer-internet company for knowledge workers
- •Doctors historically weren’t the tech ‘gatekeepers’ for their own tools
- •Free, direct-to-physician app distribution drove viral adoption
- •Leadership at major hospital systems became active users, reinforcing uptake
- 15:53 – 19:28
Patients in the loop: access to information vs misinterpretation risk
Elad raises the tension between empowering patients and avoiding misuse of complex medical evidence. Nadler supports patient agency but stresses how hard it is to interpret trials without statistical literacy; OpenEvidence’s compromise is physician-generated patient handouts and other communication artifacts.
- •Patients should have agency because medicine is imperfect
- •Clinical trial interpretation is difficult without statistical training
- •Risk: fear/hope-driven misreading leading to conflict or poor decisions
- •Common workflow: generate patient handouts, prior auth and appeal letters
- 19:28 – 22:12
Future of medicine: trust, automation, and the ‘pilots in the cockpit’ analogy
Elad asks how medicine changes over 10–20 years; Nadler argues forecasting is hard in an event-horizon moment. He predicts doctors remain in the loop largely because humans personify trust, using aviation autopilot as an analogy for why full automation can be technically feasible yet socially resisted.
- •Long-term projections are unreliable amid rapid technological change
- •Even proven automation (autoland) doesn’t eliminate human operators
- •Human trust is anthropomorphic; people want accountable agents
- •Interfaces and ‘visual surrogates’ can shift comfort but slowly
- 22:12 – 26:44
Medical education is being forced to invert by knowledge growth
Sarah asks about the shrinking half-life of medical knowledge; Nadler claims medical education must change radically. He cites fast growth in medical literature and argues continuing medical education will become the dominant mode, since most practicing knowledge is learned post-training.
- •Medical knowledge growth rates overwhelm traditional training models
- •Even conservative estimates imply impractical daily reading loads
- •CME becomes primary; medical school becomes a smaller fraction of learning
- •Top physicians report most of what they practice was learned after training
- 26:44 – 30:40
Residency reforms: evidence-based practice, curbside consults, and distributed care
Elad asks how training structures like residency will adapt. Nadler points to avant-garde programs that emphasize evidence-based (not just guideline-based) medicine, normalize curbside consults, and use interdisciplinary groups—while acknowledging cost and specialist scarcity as constraints.
- •Residency models are being updated at leading institutions
- •Curbside consults: rapid peer input for complex cases
- •Distributed, interdisciplinary decision-making improves quality but is costly
- •AI tools can approximate ‘panel’ expertise in under-resourced areas
- 30:40 – 34:12
Preventative health and consumer behavior: culture, walking, purpose, and portions
Sarah pivots to what consumers can do for prevention; Nadler uses Japan as a cultural contrast. He emphasizes simple, known behaviors—walking, sustained purpose/work, and portion control—arguing the challenge is will and culture more than discovering new checklists.
- •Prevention is often about adherence to well-known behaviors, not novelty
- •Walking culture and daily movement as a foundational health driver
- •Purpose and continued engagement later in life as a protective factor
- •Diet framed around portions and restraint rather than extreme restriction
- 34:12 – 36:02
Cultural constraints on public health messaging and behavior change
Elad asks how to break cultural patterns that prevent honest health conversations. Nadler argues the pendulum is swinging back toward more evidence-forward discourse, broadening beyond weight to other risk mitigants like cognitive engagement to reduce neurodegenerative risk.
- •Health conversations can be constrained by politics and social norms
- •Shift toward discussing evidence-based mitigants more openly
- •Example beyond weight: ongoing cognitive challenge to mitigate dementia risk
- •Small daily practices can support long-term brain health
- 36:02 – 37:27
Lessons for entrepreneurs: consumer-grade UX and speaking to users as people
Sarah asks what OpenEvidence implies for other fields. Nadler says medicine is unique, but the psychology isn’t: adoption accelerates when builders treat professionals as consumers, communicate in a novel, human way, and break through entrenched industry gatekeeping dynamics.
- •Human adoption dynamics generalize across industries
- •Treating knowledge workers as consumers can unlock growth
- •Novel framing and direct address can break industry ‘cul-de-sacs’
- •User psychology can be the true rate limiter, not tech feasibility
- 37:27 – 41:13
Rationalism vs will: where great companies really come from
The discussion turns philosophical: Nadler rejects 'if you build it, they will come' as simplistic and argues that outsized outcomes often come from will, drive, and obsession more than idea provenance. He critiques overly rationalized founder myth-making and emphasizes compulsive motivation as the differentiator.
- •Skepticism of ‘build it and they will come’ narratives
- •Western rationalist founder stories miss the role of will and compulsion
- •Motivation sources vary (proving others wrong, channeling aggression, etc.)
- •Big ideas can be obvious; execution requires extraordinary drive
- 41:13 – 44:48
Motivation and recruiting: hiring for ‘propulsion systems’ over pure IQ
Sarah asks how this lens shapes recruiting. Nadler claims intelligence correlates only moderately with output and looks for people with intrinsic drive who don’t require elaborate management systems—so leadership can focus on removing obstacles rather than manufacturing motivation.
- •Output depends on more than intelligence; drive is decisive
- •Prefer candidates with strong internal propulsion, even if unexplained
- •Over-analysis of motivation can diminish it; he avoids ‘killing the engine’
- •Hiring self-driven people reduces the need for heavy management
