PillPack Founder Parker Builds General Medicine to Sell Care as SKUs

PillPack创始人Parker创立General Medicine，将医疗护理打造为标准化可售SKU 由于人工智能让临床建议无处不在，医疗的核心瓶颈已转向行政流程推进，从而催生了将医疗程序转化为透明、可交易零售产品的开放市场的需求。 General Medicine创始人兼首席执行官TJ Parker指出，尽管生成式AI使得医疗信息普及化，但预约挂号、获取透明价格以及落实诊疗等重重阻碍，依然让患者寸步难行。为弥合这一鸿沟，General Medicine将临床诊疗流程标准化为独立的、可交易的消费级SKU，并辅以实时保险福利引擎和无限制的转诊网络提供支持。借鉴PillPack和Amazon Pharmacy的运营经验，Parker展示了消费者便利性与价格透明度如何从根本上改变患者行为，证明了医疗需求的价格弹性远高于行业传统巨头的假设。该平台最终将自营基础医疗服务与第三方市场履约相结合，实现行政负担自动化，并将临床建议转化为即时的一键式医疗服务。 General Medicine正在构建一个数字化医疗“商店”，将临床医疗程序和诊断转化为代码化、可交易的SKU，以解决获得AI建议与实际落实诊疗之间的行政瓶颈。 (0:00) General Medicine的数据显示，80%的患者表示曾因繁琐流程而推迟就医，其中30%的人因传统系统过于不便而推迟治疗至少六个月。 (10:00) Amazon Pharmacy在展示保险自付额的同时公布预付现金价，促使近50%的处方量转向现金支付——其中一半的用户为了避免保险行政流程的延误，甚至愿意主动支付更高的费用。 (33:38) 通过利用大语言模型解析福利协调（COB）文件和实时累计额度数据，General Medicine仅凭患者的驾照信息即可预先呈现精准的自付费用。 (31:18)

General Medicine CEO TJ Parker and venture investor Vineeta Agarwala argue that generative AI has solved healthcare's information gap, shifting the primary bottleneck to administrative execution. To address this friction, General Medicine translates clinical pathways, insurance rules, and billing codes into a digital storefront of transactable consumer items. The Bottleneck Shift from Information to Execution (0:00) While tools like large language models allow patients to research diagnoses and treatment options easily, algorithmic outputs often leave users with an uncoordinated administrative checklist, including clinic phone calls, diagnostic scheduling, and prescription pick-ups. Parker notes that survey data from General Medicine shows 80% of customers have postponed care due to systemic inconvenience, with 30% putting off care for at least six months. Rather than forcing patients through fragmented channels, General Medicine introduces a consumer storefront model. By pre-aggregating longitudinal medical records and insurance eligibility before visits, appointments shift away from administrative intake toward collaborative clinical decisions. The platform codifies subsequent steps—such as lab tests, imaging, or specialist appointments—into discrete digital orders, demonstrating the ability to book same-day specialist appointments and schedule urgent breast biopsies within three days. Standardizing Medical Procedures into Transactable SKUs (18:03) Building a universal healthcare store requires solving what Parker characterizes as a 10- to 15-year developmental lag behind mainstream consumer tech. While Amazon Pharmacy required Parker's earlier team to navigate 40 bespoke bilateral integrations to insulate protected health information from non-compliant infrastructure, General Medicine was constructed de novo as an open network without proprietary technical lock-in. The catalog relies on a two-step translation layer that converts clinical needs into standardized marketplace stock-keeping units (SKUs). First, foundational primitives manage prescriptions, imaging, and lab routing. Second, each SKU is paired with Clinical Intake Questionnaires (CIQs) to assess clinical indications and linked to back-end billing crosswalks that map medical coding (such as CPT differences). Standardizing procedures into universal items allows multiple providers to fulfill the same service, introducing competitive marketplace pricing analogous to e-commerce platforms. Hybrid Supply Networks and Automated Price Transparency (20:00) To solve the classic marketplace cold-start problem without contracting thousands of clinicians individually, General Medicine established its own medical group to serve as a first-party general practitioner layer. Because licensed clinicians can refer patients to any specialist, testing facility, or pharmacy nationwide, supply is not bounded by formal commercial networks. The platform couples this unconstrained routing with targeted integrations, including direct booking across labs and retail checkouts spanning over 50,000 pharmacies. Simultaneously, the platform calculates real-time out-of-pocket costs by ingesting basic identity data from a driver's license without requiring insurance card scans. Using large language models, the engine crosswalks member IDs, parses complex Coordination of Benefits (COB) documents, tracks deductible accumulator balances, and processes public machine-readable payer pricing files. Parker points out that standard scheduling obscures massive price disparities—citing a personal knee MRI that cost $1,700 at one Intermountain facility but only $600 at another under identical insurance coverage. Consumer Price Elasticity and Monetizing Operational Convenience (33:38) Parker challenges the healthcare industry assumption that consumer demand is price inelastic. Drawing from Amazon Pharmacy transaction data, simply presenting a cash price alongside an insurance copay shifted nearly 50% of prescription volume to cash—compared to a baseline retail rate of 5% to 7%. Crucially, half of those cash transitions occurred when cash was slightly more expensive (e.g., $17 cash versus a $12 copay), demonstrating that consumers actively pay small out-of-pocket premiums to bypass prior authorization delays, multi-day refill lags, and administrative paperwork. General Medicine applies this behavior to its business model, capturing transaction value across hybrid marketplace services while validating consumer willingness to pay explicit convenience fees. Paralleling platform fees on DoorDash or Uber Eats, patients accept direct costs when administrative legwork is automated, particularly when platform price transparency simultaneously uncovers hundreds of dollars in net savings across downstream procedures. General Medicine argues that structuring healthcare as an open, standardized marketplace can convert diagnostic advice into prompt clinical care. However, the model's broader systemic impact will depend on whether its administrative convenience fees and automated coordination layer can consistently absorb the operational variances of legacy health systems and non-integrated third-party providers.

General Medicine 联合创始人 TJ Parker 与投资人 Vineeta Agarwala 指出，生成式人工智能虽已消除医疗信息的获取门槛，但医疗体系的真正瓶颈已转向执行层面的行政摩擦；为此，该平台构建了一种消费级“医疗商店”，将诊疗服务拆解为可交易的标准 SKU，实现即时价格透明与端到端履约。 瓶颈转移与从建议到即时执行 (0:00) TJ Parker 指出，患者使用大语言模型查询病症固然能够获取详尽的可能诊断，但算法输出最终仍会演变成繁琐的行政待办清单，包括联系诊所预约、寻找化验中心或协调取药。General Medicine 的调查显示，80% 的受访用户曾因流程不便推迟就医，30% 的人甚至推迟长达六个月。Parker 根据早年在家族药房担任配送员的观察强调，慢性病患者用药依从性差的根源往往是冰箱贴表格、错位续方等琐碎物流障碍，消除这些摩擦本身就是关键的临床干预。因此，平台试图将医生、检验、影像和处方整合为一个类似电商的即时交易界面。 这一架构不仅用于避孕药续方或血脂化验等常规需求，也能支持复杂的重症就医路径。在面对乳房肿块等疑似恶性肿瘤的情形时，平台将影像检查与活检排期标准化，在三天内便完成了超声引导活检的协调，对比传统转诊动辄数周的延误优势显著；而在内分泌等通常需等待五到六个月的专科号源上，平台也实现了当日预约。此外，通过提前整合纵向病历与医保覆盖，看诊时间从以往耗时 10 到 15 分钟的既往史录入，转变为医患共同探讨治疗方案与即时开单。 医疗服务SKU化与双边供给架构 (18:03) 为了让医疗服务像商品一样陈列在货架上，General Medicine 将非标医疗操作拆解为标准化消费级 SKU：前端配置临床问诊问卷（CIQ）以评估临床适应证，后端则通过代码映射对照表（Billing Crosswalks）将复杂的 CPT 计费与支付编码转化为清晰标的。这种标准化使多家医疗机构能够针对同一 SKU 开展履约，通过引入第三方集市竞争机制推动服务降价并扩充供给。 针对双边平台的冷启动难题，平台采用了“第一方+第三方”的混合供给模式。General Medicine 组建了自有执业医师团队充当全科医生层，提供即时初级保健并安排专科电子会诊；同时，利用正规执业资质，平台无需等待商务签约即可向全美任何外部专科医生或线下机构分流患者。对于愿意深度打通的实验室与全美超 50,000 家零售药房，平台则通过深度 API 接入实现类似 OpenTable 的一键预约与处方取药。 即时透明定价与打破价格无弹性迷思 (31:18) 针对医疗需求缺乏价格弹性的传统观念，Parker 根据其创立 PillPack 与 Amazon Pharmacy 的运营数据予以反驳：在结算界面直接并列展示自费现金价与医保共付额后，近 50% 的处方量转向了现金结算，较传统药房 5% 至 7% 的自费比例提升近 10 倍。尤为关键的是，其中半数自费转化发生在现金比医保自付更贵的情形下（例如多付 5 美元自费以规避 12 美元共付），患者自愿支付溢价是为了绕过保险机构的事先授权拖延与分批取药阻碍。 在门诊透明定价方面，General Medicine 借助大语言模型管道，仅凭患者驾驶证信息即可自动提取医保标识符（Member ID、RxBIN 及 PCN）和福利协调文档（COB），再结合免赔额累加器状态与公开的谈判费率，在前台生成确切的个人自付价格（例如清晰的 30 美元门诊费）。Parker 援引个人膝盖核磁共振的经历补充指出，同一家健康系统在相近区域内，同一张保单下的相同检查标价可出现 1,700 美元与 600 美元的巨大反差，而即时透明化正是消除此类信息不对称的核心工具。 开放网络协同与就医便利性变现 (38:15) 相较于 Amazon Pharmacy 当年为满足 HIPAA 与健康隐私合规而在 40 个内部团队间艰难改造旧有闭环体系，General Medicine 采取从零构建的开放网络，允许患者在本地任意便利地点接受护理。平台搭载的 GPT 模型基于完整病历运行，在诊前为医生提炼病情摘要，在诊后则将结肠镜或乳房摄片等预防性筛查建议直接转化为商城内可一键下单的交易 SKU，避免传统纸质转诊的流失。 在商业模式上，Parker 认为医疗体系不存在单一超级入口，General Medicine 并非取代现有医疗系统，而是作为白标目录和交易前端赋能医院与保险机构。如同消费者愿意在 DoorDash 或 Uber Eats 上为节省精力和配送履约支付服务费一样，当平台能通过比价透明度为患者节省数百美元并解决挂号繁琐问题时，消费者对医疗服务便利性同样表现出明确的直接付费意愿。 TJ Parker 与 Vineeta Agarwala 描绘了将医疗重心从信息咨询转向履约电商的路径。但这一构想在全美推广的实际效果，仍取决于底层计费引擎应对各类复杂险种政策的准确率，以及在没有排他性排他协议的情况下，外部医疗机构能否长期维持高效承接和数据互通。

## Background 本期内容由 Vineeta Agarwala 与 TJ Parker 于 2026 年 10 月 22 日在播客及音频平台发布。节目依托 BUILD 2026 大会的现场对谈，重点探讨了 General Medicine 构建医疗“商店”（Store）的全新愿景，分析在 AI 使得医疗建议极度充裕的当下，如何解决落实就医行动与理清支付成本的新瓶颈。 ## TL;DR 节目围绕医疗行业的数字化转折展开，TJ Parker 指出当前医疗的最大痛点已从“获取医学建议”转向了“采取落地行动”。General Medicine 借鉴现代电商与即时零售（如 Instacart、Uber Eats）的双边市场架构，自建全科医疗团队并打通全美转诊与药房网络，将复杂的医疗服务重构成包含临床指征与报销映射的标准可交易 SKU。同时，团队利用大语言模型（LLM）穿透医保福利协议与公开费率文件，无需实体医保卡即可实现精确的自付价格透明化。对谈打破了“医疗缺乏价格弹性”和“患者不是真正客户”的陈旧偏见，论证了消费者完全愿意为省心省时的医疗便利性买单，并展示了未来如何以开放货架和白标目录赋能整个医疗生态。 ## Atmosphere & style 整场对谈兼具深度的行业洞察与一线实战的极客质感，节奏明快且信息密度极高。主客双方既有投资人与创始人的战略共鸣，又有身为医生与前线送药司机的真实切肤之痛，在调侃与共情之间碰撞出深刻的行业反思。听感上毫不空泛造作，既有揭秘巨头内部工程血泪史的冷峻残酷，又充满着用消费级极简体验重塑庞大沉疴系统的务实与雄心。 ## Bulletin - `00:00` General Medicine 创始人 TJ Parker 指出医疗行业瓶颈已从获取医疗建议转向执行就医行动；平台通过构建电商式“医疗商店”，提供端到端的价格透明与一键预约结算履约。 - `31:18` 团队通过 LLM 解析冗长晦涩的医保福利协调协议（COB）并结合医院公开费率文件，实现仅凭用户驾照信息即可秒级输出精准门诊自付金额（如“自付 $30”）。 - `33:23` PillPack 与 Amazon Pharmacy 数据打破“医疗价格无弹性”迷思：结账时仅并排展示现金价与医保价，便推动自费比例从传统 5%-7% 飙升至近 50%，其中 25% 的订单用户甚至自愿为避开核保授权等繁琐流程而支付现金溢价。 - `20:00` General Medicine 采取 1P 自营全科结合 3P 外部专科的双边市场架构，利用全科医生天然的法定转诊权打通全美任意医疗机构与 50,000 余家药房，避开前期沉重的专科技术集成壁垒。 ## Key takeaways - `00:00` AI 与大语言模型虽然让健康建议与初步诊断变得唾手可得，但真正的就医瓶颈转变为挂专科号、预约检查和厘清保险账单等现实落地环节。 - `02:30` 在看诊前预先聚合抓取患者全量电子病历并结构化呈现，能将以往耗在基础问卷上的前 15 分钟转变为高效的“顾问式”医患深度沟通。 - `05:00` 将复杂的重症诊疗拆解为标准化底层原语（Primitives），使多环节高焦虑的乳腺癌筛查及穿刺活检耗时从数周缩减至短短三天。 - `07:30` 医疗领域的“便利性”并非可有可无的奢侈体验，而是直接决定慢病患者服药依从性和临床预后的核心医疗支柱。 - `10:00` 医疗数字化进程落后普通消费零售 10 至 15 年，当前正在从早期电商阶段全面迈向如同 DoorDash 和 Instacart 的线上线下互联与即时履约时代。 - `11:54` Amazon Pharmacy 曾被迫对 40 个非 HIPAA 兼容的底层零售团队做艰难的定制改造，而 General Medicine 采用全新从零搭建（de novo）避免了巨头系统的历史包袱。 - `17:30` 医疗体系缺乏电商维度的通用“产品目录”，平台通过明确临床指征、动态 CIQ 问卷及后台报销代码映射（Crosswalk），将医疗服务真正转化为可比价的标准 SKU。 - `20:00` 自建第一方（1P）全科医疗团队不仅能提供当日专科电子会诊（e-consult），更能借助全科医生法定转诊权限打通全美所有第三方（3P）医疗机构与专科服务。 - `26:00` AI 架构在患者端实现基于完整病历历史的深度交互，在医生端前置合成病程摘要，并将临床预防性医嘱直接自动生成为店内可交易的标准化 SKU 项。 - `31:18` 利用大语言模型（LLM）穿透身份信息并结构化清洗医保 COB 条款，结合实时免赔额度与公开费率文件，后台自动完成复杂的计费测算并向前台输出极简的自付一口价。 - `33:23` 药房购物车并排展示价格的数据证实消费者拥有极高自主决策权，不仅大量因省钱转走自费，更有多达 25% 的用户愿意为省去繁琐核保流程直接掏溢价买单。 - `35:28` 同一医保下相隔不远的同系统医疗机构做膝部 MRI 存在 $1,700 对比 $600 的三倍荒谬差价，印证了消除医疗信息黑洞与推行价格透明的紧迫性。 - `37:44` General Medicine 采取开放商店与白标目录（White Label Catalog）模式，赋能大型医疗系统与支付方构建现代化的消费级就医货架。 - `40:00` 类似于为 DoorDash 外卖便利买单，患者高度认可专科一键预约和转诊履约的省心价值，对医疗便利性的付费意愿正在构筑全新的商业模式。 ## Detailed walkthrough ### `00:00` 瓶颈的转移：从获取医疗建议到采取实际行动 在播客开篇，讨论迅速切入了当前医疗行业最核心的痛点转变。过去，普通患者很难轻易获取专业医疗知识与诊断建议，但生成式人工智能（如 GPT）的普及彻底颠覆了这一点。现在，几乎所有人都能利用大语言模型迅速生成一份详尽的健康解答或下一步指南。然而，这份建议通常会带来一长串繁重的待办事项：患者需要自己打电话预约全科医生、挂专科门诊号、去不同的实验室做化验，或者奔波于各影像中心预约 CT 与核磁共振。一旦离开屏幕上的建议，患者立刻被拽回到传统医疗体系繁琐、低效且支离破碎的现实泥潭中。 TJ Parker 指出，医疗行业的真正瓶颈已经发生了根本性位移。正如 TJ Parker 所言：“While advice is abundant, the problem to be solved now is making it super easy to take those actions.”（尽管建议唾手可得，如今亟待解决的问题是让落实这些行动变得极其简单。）面对这一困境，General Medicine 提出的解决方案是将医疗体系重构成一个直观的“商店”（Store）。在现代生活的其他任何领域中，消费者的决策逻辑高度一致：搜索并浏览可选方案、清楚获知价格与可用时间、一键购买完成。General Medicine 致力于把这套顺畅的电商化体验引入医疗流程，让患者从获知建议到落实行动无缝衔接，消除横亘在知晓与就医之间的层层阻碍。 ```mermaid flowchart LR accTitle: 医疗瓶颈转移流程图 accDescr: 展示 AI 如何解决建议瓶颈，并将核心挑战推向行动层，最终通过商店模式解决。 subgraph 建议层 ["建议层（AI 已解决）"] A["患者利用 GPT 提问"] --> B["即时生成医疗建议与方案"] end subgraph 行动阻碍 ["行动层（当前现实瓶颈）"] B --> C["漫长预约专科与医生"] B --> D["寻找化验所与影像中心"] B --> E["医保报销与账单价格不明"] end subgraph 解决方案 ["General Medicine 商店模式"] C & D & E --> F["统一的医疗商店 Storefront"] F --> G["查价格 / 明权益 / 一键购买履约"] end ``` ### `02:30` 双峰需求与重构医患沟通体验 面对医疗场景的多样性，患者的需求呈现出鲜明的双峰分布（Bimodal）。一类是目标明确的日常刚需：例如患者清楚自己需要做血脂检查（Cholesterol panel），或者按期续订避孕药处方，这类低复杂度的需求本身无需过多医学推导；另一类则是复杂的未知病情：患者并不清楚具体该怎么办，但他们对可能的治疗路径和选择空间更加明智。无论是哪种情况，传统医疗体系中最折磨人的往往是信息孤岛。 针对这一问题，General Medicine 投入大量底层工程，在就诊前完整抓取并整合患者的全量医疗记录（Medical records），同时全面解析其医保福利（Insurance benefits）。在医生接诊前，这些结构化信息已经被清晰呈现在医生端界面上。这种数据前置从根本上改变了就医沟通的质量。过去，由于医生完全不了解患者过往病史，前10到15分钟只能用于机械地询问、录入基础信息；而在 General Medicine，这15分钟变成了真正平等的“顾问式咨询”（Consultative visit）。医患双方可以专注于探讨不同的医疗路径、比较利弊。一旦达成一致，行动流程变得像网购一样简单： - 如果确定需要验血化验，患者直接点击“购买”即可完成下单。 - 如果需要转诊专科医生，系统会即时筛选出符合其医保网络范围的所有专家名单，患者可在线直接选定并预约。 主持兼投资人 Vineeta Agarwala 对此深有感触。她回忆起自己在住院医师规范化培训期间，导师曾传授过一条被整个行业视为理所当然的经验：“Sometimes for a primary care patient, you could split their visits up into two to three visits and use the first visit just to make sure you have all the records.”（有时对于全科初诊患者，你可以把就诊拆成两到三次，第一次看诊仅仅用来确保收集齐所有病历数据。）这种受制于数据割裂而迫使患者跑三次才能看到检查结果的陈旧就医范式，正在被数据即时聚合的全新模式彻底打破。 ### `05:00` 从简单化验到重症筛查：高复杂度医疗的原语构建 对于外部将 General Medicine 误解为仅适用于低复杂度、轻症护理（Low-acuity）的看法，TJ Parker 予以了坚决反驳。他强调，团队通过将医疗全流程中的各项服务抽象并代码化为易于理解的独立单元（Primitives），不仅能轻松应对简单的胆固醇检测，更能在面对癌症筛查等高压、多环节的复杂医疗旅程中展现出关键价值。 以女性乳房自检发现肿块这一令人极度焦虑的场景为例，患者往往身陷迷茫：究竟该找哪位医生、是否需要做乳腺钼靶（Mammogram）或乳房核磁共振（MRI）、由谁来开具检查单。针对这类多节点（Multiple touchpoints）的重大健康事件，General Medicine 能够将每一个断裂的中间步骤串联起来并实现无缝衔接： - **亲历案例**：TJ 透露，一位身边亲近的人最近通过 General Medicine 进行了乳腺钼靶检查，结果异常需要尽快穿刺活检（Biopsy）。在传统医疗体系中这往往需要耗费数周的周转与协调，而 General Medicine 帮助她在三天之内成功安排并完成了穿刺活检，后续明确为良性；即便是恶性肿瘤，系统也能立刻对接后续所需的肿瘤专家并跟进整个治疗路径。 - **专科预约突破**：对于很多大型医疗集团动辄需要排队四到六个月的专科，平台同样实现了突破。Vineeta 提到她在亲身测试产品时，竟然直接约到了“当天”的内分泌科门诊（Same-day endocrinology appointment），而以往在普通医院系统通常需要等待半年之久。 TJ 总结道，正是由于把基础流程中的每一件小事执行到了极致，平台才有能力顺理成章地高效解决此类高度复杂的医疗重症问题。 ### `07:30` 运药司机的实地启示：便利性如何重塑患者健康 TJ Parker 创立 PillPack 及后来主导 Amazon Pharmacy 的初心，深植于他年轻时在新罕布什尔州为父亲经营的家庭药房工作的经历。尽管他一开始是在药房柜台担任药剂技师（Pharmacy tech），处理常规处方和发药工作，但他很快主动选择成为一名送药司机。这次视角的转变彻底改变了他对医疗问题的认知：他从柜台的防护网后走了出来，直接深入到患者真实的家庭生活当中。 在患者家中，TJ 观察到了一种惊人的一致性问题： - **家庭行政负担**：几乎每一位慢性病患者的冰箱上，都贴着一张手动填写的 Excel 表格或挂历，密密麻麻地用荧光笔做着标记，划掉已服用的药物、标注重复取药（Refill）的日期。每一个普通家庭都在为管理复杂的药物治疗历程而筋疲力尽。 - **视障用药障碍**：他曾服务过一位患有严重黄斑变性（Macular degeneration）的老年患者，视力严重受损，完全看不清药瓶上打印的小字药名。每次上门送药时，TJ 都必须拿一只粗体马克笔（Sharpie）在瓶身上写一个巨大的英文字母“M”，以便老人能分清那是美托洛尔（Metoprolol）。 这些切肤之痛让 TJ 意识到，医疗领域所谓的“便利性”（Convenience）绝不是无关痛痒的奢侈体验，而是直接影响服药依从性和临床预后的核心支柱。正如 TJ Parker 总结道：“Convenience matters a ton if you want to help people be healthy. Arguably, it is one of the most important things to solve for in a world that this category has...” 处理好日常细节中的阻力与不便，正是真正帮助人们走向健康的最有效手段。 ### `10:00` 医疗体验落后零售十余年与走向“现实连接” TJ Parker 首先指出，医疗服务流程中充斥着巨大的摩擦，这直接导致了严重的就医延误。根据 General Medicine 收集的客户反馈： - **80% 的客户** 曾因获取医疗服务过于繁琐而推迟就医； - **30% 的客户** 甚至将必要的诊疗推迟了长达 **6 个月** 之久。 因此，消除这些阻碍、让患者更轻松地获得服务，成了他们创业的核心主线。 Vineeta Agarwala 评价道，General Medicine 的产品与理念展现出一种罕见的特质：它在虚拟与现实世界之间架起了一座“后门坡道”（ramp out the back）。许多健康科技公司只愿停留在纯虚拟的封闭闭环里，而 General Medicine 敢于直面那个“险恶而庞大的现实世界”（the big bad world），将虚拟入口无缝连接至线下的实体诊所、实体药房以及需要面对面物理护理的场景。 TJ Parker 从宏观演进的角度解释了这种架构的必然性：医疗行业的数字化进程往往比其他消费零售品类落后 10 到 15 年。2010 年代的医疗健康大体相当于普通零售业的“早期电子商务时代”，而如今整个行业正全面迈入“双边市场（Marketplace）与线下真实世界深度互联”的阶段。正如 TJ Parker 所断言： > “So I think you're really seeing us bring that same experience that you'd get from an Instacart or an Uber Eats, but in a far more complex category.” ```mermaid flowchart LR accTitle: 零售电商与医疗数字化演化对比 accDescr: 展示零售品类与医疗行业在数字化发展阶段上的10至15年时间差 subgraph Retail ["零售与即时消费品类"] R1["早期电商时代"] --> R2["即时配送与线下互联时代<br>(Instacart / Uber Eats)"] end subgraph Health ["医疗健康领域 (落后10-15年)"] H1["2010年代<br>医疗电商初探"] --> H2["当下与未来<br>General Medicine 实体与数字互通"] end Retail -.->|"模式演进参考"| Health ``` --- ### `11:54` 从 PillPack 到 Amazon Pharmacy：在巨头体系中做医疗的残酷技术细节 针对 Vineeta 提出“亚马逊早已深谙如何打造货架商城，为何做处方药商城仍需收购 PillPack、且推进如此艰难”的问题，TJ Parker 揭示了背后极具反直觉的工程与组织内幕。 TJ 坦言，当初最大的认知突破其实是他们向管理层极力证明：“医疗健康的最大机会就是直接做亚马逊，根本不需要搞一套特立独行的逻辑”。团队内部在战略目标、资源投资和战略方向上高度一致，真正的噩梦完全在于底层的工程架构集成： 1. **底层架构对 PHI 与 HIPAA 的天然排斥**：亚马逊庞大的零售系统最初从未设想过需要处理受保护的健康信息（Protected Health Information, PHI），也完全不是按照 HIPAA 合规规范来构建的。 2. **乐观预期的破灭**：团队原本以为在 PillPack 已经做过全套合规系统，只需在原有后端上搭一层前端体验即可，“6 个月内就能上线”。但当真正深入下去时发现，要让医药体验像亚马逊原生服务一样顺畅，必须贯穿其庞大的内部基础设施。 3. **牵扯 40 个内部团队的定制改造**：为了推出最简版本（MVP）的 Amazon Pharmacy，团队必须与账户身份（Identity）、支付（Payments）、税务（Tax）等 **40 个不同的底层团队** 进行定制化对接。这些团队自身严禁触碰任何 PHI 数据，每一个微小的触点都需要极其精巧的隔离与技术方案。 正如 TJ Parker 所总结： > “The punchline is that we had to integrate into an ecosystem that was never intended to deal with PHI or be HIPAA compliant.” 这项集成工作最终一次性打通了亚马逊底层的医疗通路，但其开发代价极为惨重。 相比之下，General Medicine 采取了完全不同的全新构筑（de novo）路线： - **从零搭建免于包袱**：无需在复杂的庞大旧系统上做妥协集成，开发与迭代速度呈数量级提升； - **不被单一实体捆绑**：过去做 PillPack，团队既是软件方也是唯一药房，必须自建所有的履约基建。而 General Medicine 致力于成为一个连接任意医生、任意医疗服务商、任意药房的开放网络，患者可以在任意最便捷的线下节点接受服务。 --- ### `15:00` 什么是“综合商店”？医疗服务产品目录（Product Catalog）的创立 Vineeta 用经典的“杂货店（General Store）”作比喻：杂货店之所以成立，核心在于“它必须包罗万象”，绝不能只卖两排货架的商品（就像《宋飞正传》里乔治·斯坦布伦纳调侃的 Art Vandelay 虚构进出口生意一样）。若在患者最需要关头冷冰冰地表示“我们不提供此服务，也无法代为开具”，平台价值便不复存在。 TJ Parker 指出，医疗领域一直缺乏真正的“产品目录（Product Catalog）”机制： - 药房领域相对规范，有明确的药品统一编码（NDC, National Drug Code），商品实体定义明晰； - 但在绝大多数医疗服务中，行业仅有大量用于医保报销（Billing）和流程结算的内部代码，从未有人尝试将“医疗体系的底层逻辑”翻译成“消费者所理解的产品语言”。 在此背景下，Vineeta 调侃 TJ 重新全职出任 General Medicine CEO 的决定，戏称这是他的“Travis Kalanick（优步创始人）式回归”。TJ 坦率回应，虽然此前作为联合创始人及投资人始终在参与，但选择全面回归日常驾驶位的原因非常简单： > “I mean, it's super simple. It's more fun. It's so enjoyable to be able to be so deep in it and building the thing that the three of us, and to some degree the broader team, has wanted to exist for so long.” 随着业务高速扩展，多项具有探索性的战略布局开始开花结果，他被彻底“拉回”了一线战场。团队甚至会为了内部有人说出零售术语“商品组合（GUMIX）”而欢呼击掌，彻底被“商城化逻辑（Store-pilled）”所浸润。 --- ### `17:30` 医疗 SKU 的工程化机制：临床标准、CIQ 与报销代码映射 TJ 深入拆解了 General Medicine 如何在医疗领域构建起一个拥有极高丰富度的商品目录，将其细化为类似电商的“SKU”机制。这一系统建立在两层核心逻辑之上： #### 1. 基础设施图元（Primitives）的完备性 要提供海量品类，首先必须具备支持各类医疗旅程的基础能力：开具化验单（Labs）、处方用药（Prescribing medications）、转诊专科医生（Specialist referral）等。缺乏这些基础图元，就无法支撑复杂的健康诉求。 #### 2. 将非标医疗转化为严谨的 SKU 每个上线商城的 SKU（例如“肩部疼痛诊治”、“某项 MRI”、“全套血液化验”）都必须具备以下结构： - **明确的临床标准与 CIQ（临床问卷）**：每个 SKU 对应一套定义严谨的临床指征。系统通过特定的临床摄入问卷（Clinical Intake Questionnaire, CIQ）收集患者病况，判断该项服务是否适用。即使面对同一种药品（例如使用 β-受体阻滞剂来治疗焦虑症还是高血压），其对应的临床标准和 CIQ 也完全不同； - **底层报销代码的映射交叉（Crosswalk）**：患者眼中的同类服务，在后端报销时可能对应截然不同的代码与账单逻辑。例如“肠镜检查 A”和“肠镜检查 B”，在消费者看来都是做肠镜，但在系统底层由于检查目的（预防性筛查 vs 诊断性活检）不同，费用结算与报销路径差异巨大。平台必须承担翻译与映射的工作，对齐患者认知与后台账单； - **多供给端接入与价格竞争**：定义标准 SKU 后，平台能够让成百上千个具备资质的外部医疗服务提供方（Providers）在同一个标准化商品项下列出服务。 正如 TJ Parker 所阐述的电商核心红利： > “The benefit is 2,000 sellers can all sell the same thing. What that does for customers is it drives the price down and it creates competition.” 通过将医疗服务产品化、标准化，并引入广泛的履约方竞价，平台才真正有希望打破传统医疗价格黑盒，实现压降成本与透明化消费。 ```mermaid flowchart TD accTitle: 医疗服务向标准SKU转化的机制图 accDescr: 描述如何将临床准入、CIQ问卷与后台报销代码对齐为标准电商SKU subgraph UserView ["消费者视角"] A["医疗需求 / 商品展示<br>(如: 肠镜检查 / 肩部疼痛)"] end subgraph Engine ["SKU 转化与匹配中枢"] B["临床准入标准匹配"] C["动态 CIQ 问卷摄入<br>(按具体病因与适应症分流)"] D["报销代码交叉映射 (Crosswalk)<br>(将诊断代码/计费代码转换为标准账单)"] end subgraph Fulfillment ["多方履约生态"] E["服务提供方 A (报价/接单)"] F["服务提供方 B (报价/接单)"] G["服务提供方 C (报价/接单)"] end A --> B B --> C C --> D D --> E D --> F D --> G ``` ### `20:00` 医疗“商品目录”与自建第一方供给的双边市场策略 在医疗消费场景中，患者往往无法直接界定自己的需求所对应的专业供给。TJ Parker 指出，即使同属于皮肤科，一个专门做痣筛查（mole check）的医生可能并不愿意处理其他繁杂的皮肤问题，因此平台必须精准地将消费者与正确的提供方相匹配。商品目录（product catalog）的本质，正是一个“翻译层”——在“消费者想要什么”与“哪些人能提供且会以何种特定方式计费”之间架起桥梁。针对 Vineeta Agarwala 提出的双边市场冷启动难题（是先验证需求还是先签约供给？是否需要挨家挨户敲开全美每位心血管医生的门？），TJ 阐述了 General Medicine 的核心破局点： > “从战略角度来看，我们遵循了那些最成功的‘第一方与第三方混合型（hybrid 1P/3P）’市场的成长路径。”（TJ Parker） 团队早早押注必须优先获取消费者并为其提供极宽维度的服务。为了避免因供给不足而受限，General Medicine 并没有盲目去线下地推签约医生，而是决定自建医疗服务团队（Medical Group）。这一自建队伍实质上承担了全科医生（GP）的职责，能够覆盖广泛的基础医疗需求，并以极高效率处理专科会诊。例如，平台能为患者提供当天完成的专科电子会诊（e-consult），即便后续仍需转入线下就诊，也能极大缩减等待周期。 ```mermaid flowchart TD accTitle: 混合型双边市场与全科转诊架构 accDescr: 展示自建第一方全科医生与外部第三方全美专科医生的连接机制 subgraph Demand["需求端：消费者"] C["患者需求与历史病历"] end subgraph Platform["General Medicine 平台 (翻译层与商品目录)"] Catalog["SKU 商品目录<br/>(标准化服务与透明定价)"] GP["第一方自建医疗团队<br/>(承担 GP/全科中枢职能)"] end subgraph Supply["供给端：全美医疗网络"] EConsult["专科线上同日 E-consult"] InPerson["全美任意线下专科医生<br/>(基于全科转诊权，无需前置深度系统集成)"] Pharmacies["全美 50,000+ 家药房结算网络"] Labs["深度集成化预约 (类 OpenTable 体验)"] end C --> Catalog Catalog --> GP GP --> EConsult GP --> InPerson Catalog --> Pharmacies Catalog --> Labs ``` --- ### `22:00` 突破系统集成壁垒：利用全科转诊权力打通全美供给 在供给拓展上，General Medicine 采取了一项打破行业惯例的创新：平台能够将患者转诊至全美任何一家医疗机构或任何服务类别，而完全不以该机构是否入驻其平台为前提。这一机制依赖于医疗体制中全科医生的天然法定属性——全科医生（GP）拥有将患者推荐至任何符合临床指征的执业医生的权力。TJ Parker 强调，团队绝不会因为技术集成深浅或商业合作关系而人为限制患者可获得的照护范围。 Vineeta Agarwala 对此进行了深入剖析：与必须决定是否将洗发水铺进 Costco 的日用品供应商截然不同，专科医生在医疗体系中长期依赖转诊流入。由于专科医生往往并非处于“需求受限”状态，高质量的转诊实质上是在为他们提供一种“零获客成本（zero-CAC）”的精准患者获取渠道。针对这一供给特性，TJ 解释了平台的双轨制逻辑： - **不设人为门槛**：绝不因尚未打通 API 而剥夺患者选择权，通过全科转诊底座实现全美供给可达。 - **特定垂直领域深度切入**：例如在药房网络上，General Medicine 几个月前上线了直结功能，支持患者直接在应用内完成结算并在全美 50,000 多家药房自提或配送，给消费者带来极致顺畅的“深度集成”体验。 --- ### `24:30` 供给端合作的“Yes, and...”策略与无缝就诊体验 当谈及未来与大型医疗系统（Health Systems）及远程医疗（Telemedicine）机构的正式合作时，TJ Parker 用“Yes, and...”来形容当前策略： - **推进类“OpenTable”式的无缝预约**：尽管无需合作协议也能完成传统转诊，但深度技术集成能将体验升级为像 OpenTable 订座一样的即时预约。目前该体验已在检验化验（Labs）场景落地，省去了人工介入和繁琐的中间跟进。 - **与大型医疗系统深度协同**：通过前置系统规则与准入标准（criteria）匹配，帮助符合特定指征的患者直接进入手术或诊疗程序，绕过漫长无序的前期检查（workup）。 - **聚合远程医疗服务生态**：联合各大外部远程医疗团队，平台可以根据各家机构针对不同支付方（Payers）的挂号自付额（copay）结构进行智能推荐，确保终端用户获得性价比最高、流程最顺畅的选择。 > “这些集成对我们而言，核心动力始终在于：如何帮助你获得更好、更快、更无缝的医疗准入；我们正在医疗系统与远程医疗两端全力推进。”（TJ Parker） --- ### `26:00` 双向闭环的 AI 引擎：全量病历驱动与可交易 SKU 推荐 Vineeta 提到，如今在许多肿瘤专科门诊，新确诊的结肠癌患者走进诊室时，已经通过 AI 了解到自己很可能需要接受 FOLFOX 化疗方案，甚至会主动向医生询问何时开始化疗。患者与医生的沟通跳过了基础科普的“第 0 步”，直接进入了更具深度的决策讨论。TJ 透露，General Medicine 的 AI 架构正是为了放大这种高效沟通，并形成双向闭环： 1. **患者侧的全病历 AI 对话**：普通消费级 AI 工具往往因缺乏患者完整的医疗数据导致信息密度极低。而在 General Medicine，患者的交互式 GPT 拥有其完整的全量电子病历上下文（total medical record），患者在就诊前就能依托完整历史了解各种选项。 2. **医生侧的就诊前置合成**：平台底层将患者的完整病历与诉求进行结构化提炼，在医生接诊那一刻，即时呈现患者当前病程处于哪个节点，使医患能在更高认知层面上展开深度对话。 3. **闭环持续追踪**：就诊结束后，患者可以继续与 GPT 进行后续互动，这些新的对话内容会被再次沉淀、分析，并自动汇总为医生下一次复诊的背景资料。 ```mermaid flowchart LR accTitle: AI 双向协同与自动化可交易 SKU 生成 accDescr: 展示患者、医生与底层系统之间通过 AI 实现闭环并生成直接可交易服务的机制 subgraph PatientLoop["患者端体验"] P["患者"] <-->|"全量病历支撑的深度问答"| PGPT["患者端 GPT"] end subgraph ProviderLoop["医生端工作流"] Dr["医生"] <-->|"就诊状态自动摘要与指征建议"| DGPT["临床端 GPT"] end subgraph CoreEngine["底层交易与产品层"] System["底层系统 (Primitives)"] SKU["生成可即时交易的 SKU<br/>(处方药 / 乳腺钼靶 / 肠镜)"] end P -->|"到诊接诊"| Dr PGPT -->|"就诊前上下文同步"| System System -->|"结构化病程输入"| DGPT Dr -->|"下达医嘱与预防决策"| System System -->|"自动转化为店内下单项"| SKU SKU -->|"就诊后一键履约"| P ``` 更为关键的是将临床建议无缝转化为商业目录里的“可交易 SKU（transactable SKU）”。TJ 举例说明：如果一位 55 岁未做过乳腺钼靶检查（mammography）或 50 岁从未做过肠镜检查（colonoscopy）的患者因为配药来访，临床 GPT 会自动在医生工作台中提示这些预防性照护项目。与传统医疗中医生仅口头叮嘱“你顺便去做个检查”或写一张纸质转诊单不同，系统会将这些推荐建议直接转换为患者端可以一键点击预约、结算的标准化 SKU。“在同一个商店内，所有环节一站式完成。” --- ### `28:30` 解放临床医生：将精力从药物报销目录中抽离 在传统诊疗流程中，医生往往被迫耗费大量时间去充当医保精算师和报销审核员。以 GLP-1 类减重降糖药物为例，不同商业保险的优先药物报销目录（preferred drug list / formulary）千差万别，患者的核心诉求仅仅是“拿到那款医保能报销的药”。TJ 指出，过去的系统极其低效，医生被卷入查验某款药是否在报销目录内、自付比例多少的泥潭中。 General Medicine 通过技术工具将这部分繁重的决策支持剥离，直接赋能给患者本身，让其清楚了解覆盖情况与价格梯次；医生则彻底解脱出来，将精力聚焦于核心临床评估与纯粹的医疗建议（例如探讨已报销药物与未报销药物之间的临床疗效差异）。Vineeta 总结道，这种平台模式赋予了医生强大的工具支撑，并为双边市场带来了新的进化空间——未来甚至可以基于医生对数字化工具的采用度与服务效率，实现更精细化的网络评级与激励。 ### `30:00` 医疗“商店”的精准供给匹配与临床导购 在患者寻找医疗服务的过程中，供给端的智能匹配至关重要。现实中，每个人几乎都经历过为了找对医生而在三四个专家之间辗转奔波的痛苦。Vineeta Agarwala 抛出问题：General Medicine 在筛选整理医疗“商店”的货架、并向患者精准推荐医疗服务与产品时，究竟扮演什么角色？ TJ Parker 解释说，在任意给定的就医时刻，患者衡量的维度都是多元的，首要维度往往是便利度与响应速度。例如做活检（Biopsy），速度是第一要务，而市场上提供活检操作的医疗机构非常广泛，它们并不一定非要与患者最终接受长期治疗的医院绑定。在这种情况下，患者应当优先优化“响应速度”；但如果患者确诊了某种罕见的癌症，核心诉求就会瞬间转变为“必须找到这个领域最顶尖的那位专家”。不同场景决定了截然不同的决策考量。General Medicine 的定位是在这一“导购体验”中保持极高智能，帮助用户厘清当前究竟处于哪种场景。TJ Parker 强调这绝不会让消费者孤立无援： > “And again, that isn't just the consumer on their own, like, I hope I can figure this out. We will play the role of helping you figure that out as clinicians.” 平台的临床团队会协同介入，让复杂的医疗框架和患者关心的多维因素化繁为简，协助患者与医生共同做出最明智的决策。 --- ### `31:18` 拆解支付黑盒：用 LLM 实现无需保卡的即时价格透明化 General Medicine 核心基础设施中最具颠覆性的一环，是对患者保险覆盖范围与真实自付价格的彻底透明化。Vineeta 分享了令她惊艳的亲身体验：她仅仅输入了自己的驾照信息，General Medicine 就自动把她的医保报销细节查得一清二楚，完全不需要像传统就诊那样拿出实体医保卡拍照上传。 ```mermaid flowchart TD accTitle: 医疗保险与自付价格智能解析流程 accDescr: 展示从用户身份信息输入到最终展示简单门诊价格的端到端技术链路 subgraph UserInput["用户端输入"] A["用户提供基础人口统计学信息<br/>（如驾驶证信息）"] end subgraph DataEnrichment["底层数据穿透与结构化"] B["映射匹配医保信息<br/>（Member ID / BIN / PCN）"] C["调取完整福利协调协议<br/>（Full COB）"] D["LLM 深度解析 COB 文本<br/>转化为结构化权益与自付规则"] E["拉取实时免赔额<br/>（Deductible）与自付上限进度"] F["结合公开费率文件<br/>（Open Pricing Files）测算机构报销额"] end subgraph CustomerPresentation["前端呈现"] G["统一清洗合成极简账单<br/>『本次门诊自付 $30』"] end A --> B B --> C C --> D D --> E E --> F F --> G ``` TJ 透露，这套系统在五年前根本不可能建成，其核心完全依赖 LLM（大语言模型）的突破。底层技术链路包括： 1. **身份信息交叉映射**：根据人口统计学信息瞬间跨库检索出医保身份（不仅限于卡面上的 Member ID、BIN 和 PCN）； 2. **福利协调协议（COB）的 LLM 结构化**：抓取完整的福利协调协议（Coordination of Benefits, COB），利用 LLM 对冗长晦涩的医保条款文本进行清洗并转为结构化数据，彻底弄清患者的报销资格、定额挂号费（Copay）结构及自付费用（Out-of-pocket）结构； 3. **实时抵扣与费率合并**：拉取用户当前个性化的免赔额（Deductible）已消耗状态与年度自付最高限额状态，同时抓取医院公开费率文件（Open Pricing Files），预判该服务在特定机构的真实报销额度； 4. **极简终端输出**：后台完成这一系列极其复杂的计算（TJ 戏称为“三周半跳加转体”的 Triple Lindy 操作）后，前端呈现给用户的只是极为清爽的一句话——“去这家诊所做检查只需 30 美元”。 TJ 概括这种产品体验的终极追求： > “And in many ways, I think that's the... the vibe that we want for our customers is like, we just got this. We figured it out for you.” --- ### `33:23` 戳破“医疗价格无弹性”迷思：PillPack 经验与 10 倍自费迁移 针对医疗行业常年存在的一种观点——“医疗属于刚需，患者在生病时别无选择，因此医疗需求天然缺乏价格弹性，价格透明化毫无意义”，TJ 毫不客气地反驳这纯属无稽之谈： > “Yeah, I think that's nuts, generally speaking. Obviously, inpatient, different thing. But the vast majority of medicine, I think that's kind of a crazy point of view to have.” 为了佐证价格透明对消费者行为的巨大改变，TJ 回顾了当年创办 PillPack 及主导 Amazon Pharmacy 时的经典真实数据：当时团队仅仅是在结算购物车结账页做了一个最基础的改动——将低价自费现金价（Cash Price）与医保报销价并排展示，甚至没有做任何复杂的算法优化。结果令人震惊： - **自费比例暴增 10 倍**：传统零售药房通过自费、现金或 GoodRx 折扣码结算的比例仅为 5%–7%，而他们将近一半（接近 50%）的订单从走医保直接转向了走自费现金。 - **25% 的场景绝对省钱**：在这半数选择自费的用户中，有 25% 的情况是现金自费确实比走医保 Copay 更便宜，省钱驱动用户选择自费理所当然。 - **另 25% 的场景用溢价买顺畅体验**：令人意想不到的是剩余的 25%——在这些订单中，自费现金价其实比医保报销价还要贵上几美元（比如自费要 17 美元，走医保只要 12 美元；或者自费 21 美元对比医保 18 美元）。然而用户依然毫不犹豫地选择自费直接付款。 原因在于医保报销往往伴随着繁重折磨的“悲伤路径”（Sad Path）：患者不需要再为了两款药的续方周期对不上而硬等四天，也不需要到处打电话去追核保前置授权（Prior Authorization）。用户愿意为“今天就能拿到药且省去所有扯皮推诿”多掏 5 到 6 美元。这充分证明了消费者在面对简单明了的选择时，展现出了极其强大的自主决策力。 --- ### `35:28` 打破“终端患者不是客户”陈词滥调与荒谬的检查差价 医疗行业长期充斥着另一句老生常谈：“患者根本不是真正的客户，雇主、联邦政府和保险支付方（Payer）才是买单的客户，因此被保险人永远不可能主导决策。”TJ 同样驳斥了这种说辞的荒谬逻辑：如果患者不是客户，那为什么在药房结账面对并列价格时，他们有一半的时间会主动掏出现金走自费通道？当他们选现金时是客户，选医保时就突然变成不是客户了吗？这在逻辑上根本说不通。过去二十年里，支付方投入了海量心力设计精密的定价机制，试图引导患者去往“成本最低、质量最高”的医疗机构，但从未花力气把这些逻辑清晰地传达给患者。General Medicine 的核心使命就是充当这个沟通者，消除信息不对称。 随后两人聊起更高客单价医疗服务的价格黑洞。Vineeta 提到：“12 美元对 17 美元看似无关紧要，但等你尝试预约一次 MRI（核磁共振）或超声心动图时，情况就完全不同了。”TJ 随即分享了自己不久前的亲身遭遇：他在春季与 Elliot 结伴在阿拉斯加滑雪时摔断了前交叉韧带（ACL）， Vineeta 打趣追问他怎么有空去，TJ 调侃道：“那是我还在做 VC 的时候，所以那时我还能去滑雪。”为了检查膝盖，他去预约 MRI： - 在犹他州帕克城（Park City）的 Intermountain 医疗机构做膝部 MRI，使用完全相同的医保，报价为 **1,700 美元**； - 在盐湖城（Salt Lake）同属 Intermountain 系统的医疗机构做完全一样的检查，同样医保下自付只需 **600 美元**。 相隔不远、同一系统、同等保险，差价高达近三倍，而这些信息在传统就诊体系中对普通大众是完全隐形的。TJ 直言每天无数患者都在两眼一抹黑的情况下做着巨额医疗支出决定，这种荒唐现状必须被改变。 --- ### `37:44` 医疗生态竞合：开放商店与白标目录赋能大机构 鉴于医院系统和商业保险公司长期以来也试图解决就医入口和价格透明的问题，Vineeta 提问：General Medicine 如何与这些现有行业巨头（Incumbents）展开合作？ TJ 阐明了 General Medicine 的定位：他们擅长的是面向终端用户的客户体验建设，充当传统医疗机构与现代消费者习惯之间的“翻译层”（Translation Layer）。同时，团队坚信未来绝不可能存在一个所谓的“单一就医总入口”（Singular Front Door），人们会通过形形色色的渠道寻求医疗。General Medicine 正在推进开放合作模式，赋能生态内的其他参与方建立属于他们自己的医疗服务商店： - **大型医疗系统**：拥有无与伦比的临床诊疗实力，但其挂号接入流程极其繁琐劝退。General Medicine 可以成为其平滑获客与导流的优质数字化入口； - **大型零售巨头与支付方**：许多零售企业希望建立自己的健康服务体验，大型保险公司则希望让数百万会员真正看懂各项福利条款。General Medicine 可以提供类似“白标目录”（White Label Catalog）的底层产品货架与交易能力。 关于商业模式与变现方式，TJ 指出 General Medicine 最直接的对标参照物就是极其成功的一方（1P）自营结合三方（3P）撮合的平台型交易市场（Marketplace）。虽然医疗健康领域涉及大量的合规复杂性，但本质依然是围绕医疗产品与服务的交易流转来构建自我造血的商业闭环。 ### `40:00` 为医疗便利性买单与构建“一键搞定”的医疗体验 TJ Parker 深入探讨了 General Medicine 的商业化模式与患者付费意愿背后的底层逻辑。他指出，医疗领域正在复刻外卖与叫车平台的消费习惯：“I think the other thing that we're seeing is similar to consumers' willingness to pay for convenience on DoorDash or on Uber Eats. If you can facilitate a booking with a specialist and provide the referral, they're happy to pay for that convenience.”（我认为我们看到的另一件事是，类似于消费者在 DoorDash 或 Uber Eats 上愿意为便利性付费，如果你能帮他们搞定专科医生的预约并提供转诊单，他们非常乐意为这种便利买单。）长期以来，医疗行业一直忽视了便利本身的巨大经济价值，但实践表明，患者完全愿意为了简单、可及性和省心直接掏腰包，这正在成为医疗商业模式中最具颠覆性的一点。 Vineeta Agarwala 对此高度认同，并补充强调，当平台在提供便利的同时还能带来价格透明度时，这种价值主张会变得更加势不可挡。如果平台在预约过程中消除了信息不对称，顺便帮患者省下了几百美元的看病费用，消费者为这种高效服务付费的动力就会更足。现实中，绝大多数人面对繁琐的就医流程都会陷入焦虑——手里有一堆待办事项，需要到处打电话预约、协调转诊、核对保险，耗费大量心力。Vineeta 对 TJ 所勾勒的未来充满期待，在那样的场景里，就诊流程不再是一道道繁琐的难关，而是真正实现“一键搞定”。对谈在对双方未来合作的期待以及对革新就医体验的憧憬中圆满结束。 - **消费者付费意愿转移**：对标 DoorDash 和 Uber Eats 等消费级互联网产品，医疗用户愿意直接为“买时间、省心力”的便利服务付费。 - **专科预约与转诊闭环**：平台若能一站式搞定专科医生（Specialist）预约并出具转诊单（Referral），即可直接为患者创造极高的确定性价值。 - **便利性与透明度的乘数效应**：在解决预约痛点的同时提供透明的比价机制，帮助患者节省实际就医成本，将形成强大的用户黏性与独特的盈利模式。 ## Notable quotes > `00:00` While advice is abundant, the problem to be solved now is making it super easy to take those actions. — TJ Parker > `07:30` Convenience matters a ton if you want to help people be healthy. Arguably, it is one of the most important things to solve for in a world that this category has... — TJ Parker ## Mentioned General Medicine, TJ Parker, Vineeta Agarwala, PillPack, Amazon Pharmacy, Amazon, Instacart, Uber Eats, DoorDash, Costco, GoodRx, Intermountain Healthcare, Travis Kalanick, OpenTable, GPT, HIPAA ## Video description AI is making healthcare advice abundant. But what happens when you actually need to act on it? General Medicine CEO TJ Parker argues that the new bottleneck is getting the lab, booking the specialist, filling the prescription, or understanding what any of it will cost. At BUILD 2026, a16z General Partner Vineeta Agarwala sits down with TJ to unpack General Medicine's vision for a healthcare “store”: one place where patients can understand their options, prices, and insurance coverage, then actually get care. They discuss why convenience can materially change patient behavior, how AI can connect medical records and benefits to the right next step, and what TJ learned building PillPack and Amazon Pharmacy about making a complicated healthcare system feel simple. [Source](https://mgln.ai/e/1344/afp-848985-injected.calisto.simplecastaudio.com/d1df597a-30d8-4cae-bb4c-22a8ec496e6d/episodes/ac91e2a9-5f13-4b1f-9f7e-dc9303854ed5/audio/128/default.mp3?aid=rss_feed&awCollectionId=d1df597a-30d8-4cae-bb4c-22a8ec496e6d&awEpisodeId=ac91e2a9-5f13-4b1f-9f7e-dc9303854ed5&feed=BXDamaKF)

[0:00] TJ Parker: While advice is abundant, the problem to be solved now is making it super easy to take those actions. [7:30] TJ Parker: Convenience matters a ton if you want to help people be healthy. Arguably, it is one of the most important things to solve for in a world that this category has...

[0:00] TJ Parker: While advice is abundant, the problem to be solved now is making it super easy to take those actions. [6:55] TJ Parker: Convenience matters a ton if you want to help people be healthy. [3:04] TJ Parker: Instead of spending 10 or 15 minutes in a visit trying to just download what's going on, you spend 15 minutes debating all the different options available to you.

[0:00] Convenience matters a ton if you want to help people be healthy. While advice is abundant, the problem to be solved now is making it super easy to take those actions. And we've, for a very long time, felt like the best modality for that is a store. Amazon kind of figured out how to make a store, but they kind of needed some of the insights of PillPack to really launch the pharmacy store. The practically difficult things about building Amazon Pharmacy are pretty unintuitive. What would general medicine do for a patient who feels a lump in their breast and knows they need care? Someone that's close to me just went through this. General medicine was able to help schedule biopsy within three days, get that biopsy done, and all the way down that journey, helping you access the care that you need. Some people argue that price transparency doesn't matter in health care because I don't really have a choice if I need health care. I think that's nuts. We've made the bet that... But advice only gets you so far if you still have to spend weeks figuring out how to act on it. At Build 2026, Vinita Agarwala sat down with General Medicine CEO and co-founder TJ Parker to talk about why he believes healthcare needs a store. TJ explains how General Medicine is bringing doctors, labs, imaging, specialists, prescriptions, insurance, benefits, and pricing into one experience— so patients can move from knowing what they need to actually getting it. We also discuss what he learned building PillPack at Amazon Pharmacy, why convenience can be critical to better care, and how AI could finally make the complexity of healthcare invisible to the patient. TJ, you've said that AI is making advice easier to get, but what about that action layer? Why does healthcare need a store? You know, I think we're all seeing that advice is becoming really abundant and getting to a list of answers with a GPT has become a thing that almost everyone is doing now in health care. But I think that answer almost always comes back to a set of tasks for you to go do. It might be, you should go call and book an appointment with a doctor, call and book an appointment with a specialist, go get these sets of labs or go get this imaging done. And while advice is abundant, you then are immediately stuck back with the realities of the current healthcare system. And so it has become very clear to us that the problem to be solved now is making it super easy to take those actions and to immediately go and we've for a very long time felt like the best modality for that is a store, right? Everything else in your life, you search and browse the options available to you, you understand what they cost, you understand when they're available, and you go buy the thing. And what we're trying to achieve is make healthcare that simple rather than this very complex, convoluted process we have today. And so I think In many ways, the time is right for something like this in healthcare because the bottleneck is no longer advice. The bottleneck is taking action on that advice. And it used to be the case that we thought patients didn't really know what they needed and they would never be able to find out. But that's exactly what AI has changed. So it's almost all the more frustrating to not be able to take action on something when you know what you need. What have you heard from patients? dental medicine patients on that front. Yeah, I mean, I think it's sort of bimodal, right? I think there's so many of these scenarios where customers do know exactly what they need. It's not that complicated to know you need a cholesterol panel. It's not that complicated to know you need a renewal of your birth control. This stuff is really simple. And then there's a bunch of times where they don't know what they need, but they're much more informed about the spectrum of options available to them. The most interesting thing we've heard from customers since we launched is that because we've worked so hard to collate their full medical record, we fully understand their benefits and what's available to them. And that's all presented to the physician specifically. prior to the visit, that visit is just so much more powerful. So if you want advice from a clinician, it's not that that advice is not available to you. In fact, the advice really becomes consultative. And instead of spending 10 or 15 minutes in a visit trying to just download what's going on, you spend 15 minutes debating all the different options available to you. And then once you've landed on an option with that provider, you can just immediately go take action. So if it's a set of labs, you just click buy. If it's a referral, you see all the specialists in with your benefit and you just book the referral. And so really trying to make this not just a spearfishing exercise, which I think, again, is very valuable for a certain set of use cases. But also when you want that advice, it's readily available and you can immediately take action on it. One of the things I remember a mentor in residency teaching me is that Sometimes for a primary care patient, You could split their visits up into two to three visits and use the first visit just to make sure you have all the records. And I still remember that someone gave me that advice and it was normal. It was like, yeah, because I really want to make sure I have the right data. And make sure you guide the patient and tell them that. [5:00] We're going to focus on getting the data. And then next time we're going to talk about... Might even get to revisit in the second visit. And then on the third visit, we're going to talk about what the test results were. So you're really shifting a longstanding paradigm that was blocked. by the inability to have data at your fingertips, knowledge at your fingertips, General medicine's ability to collate medical record data so seamlessly was one of the things that really drew us in. Everybody that tried the product. felt like, whew, that was magic. You got everything about me. But let's talk about the types of care in the store. that actually people really struggle to get access to sometimes in the real world. And a cholesterol test, I agree, still not easy to get, but There are a couple of places you could get one. An example kind of close to my heart are patients who go through cancer screening journeys, and you feel a lump in your chest, to get a mammogram, to get into a doctor, get the testing order? Do I need an MRI? These are all hard questions and they're happening at a time when you're pretty stressed. What would general medicine do for a patient who feels a lump in their breast and knows they need care but may not know exactly what they need? Yeah, I think this is where, because we've invested so much in what we view as kind of the primitives of the store, which is taking all of this different selection, turned it into like codified, very easy to understand, like individual experiences, right? I think once you have those primitives, A lot of these journeys are, there's multiple touch points, there's multiple interventions that are required, but each one of those touch points can become seamless. And so in that scenario, like very clearly you need to get imaging. And if that imaging is already had, so I had someone that's close to me that just went through this with general medicine, already had the mammogram and the mammogram came back as you need to go get a biopsy. And so general medicine was able to help her schedule that biopsy within three days, get it was not cancerous, but in the event that it is, we're then helping you get to the specialist that's going to need to deal with that and all the way down that journey, helping you access the care that you need. And so I think one of the I think, assumptions about general medicine as we're focused on this really simple, low-acuity stuff. It's quite the opposite. We are dealing with a number of... situations where customers have no idea what to do next, and they don't know where to go, and they can't see a specialist for four months, they don't know how to get access to that biopsy. And so I think by doing the easy stuff so well, it gives us the opportunity to also do this really complex stuff quite well. When I was testing out the product, which obviously we did a lot of, I got a same-day endocrinology appointment. And it takes six months to get an endocrinology appointment at most health systems. So let's go back to your kind of starting journey in healthcare. You grew up delivering medications for your dad's pharmacy in New Hampshire. It's a wonderful life, but I never got the backhand. What did you learn about your customers, patients, people? that you probably wouldn't have learned if you just stayed behind the counter or been behind a spreadsheet. Yeah, I've always viewed this as one of the most formative experiences for me. And I did start by working behind the counter. I was a pharmacy tech and did all the things you'd expect at your local family pharmacy. But then pretty quickly I decided I wanted to be the delivery driver. And now I'm inside of the homes of folks that are customers, which is very different than being behind the counter. And... The thing that stood out to me the most in that experience was just the consistency of the frustrations and problems that existed. It wasn't this super wide spectrum and there was a lot of variability. You saw just the same patterns of challenges that everybody had in that process. Everyone had this Excel spreadsheet on their fridge and had all their medications crossed out and highlighted and refill dates on the calendar and just like this administrative thing they were all trying to manage. trying to keep on track of everything. And that's part of what led me to start Popak, was just feeling like, you can make this way more convenient for customers. And honestly, the PillPack thing was very much, in its early incarnation, purely about convenience. And then there obviously were more interesting anecdotes. Like I had one customer that had really bad macular degeneration, couldn't read anything. Couldn't read the name of the medication on the pill bottle. Every time I dropped off the meds, I'd write a big M for metoprolol and Sharpie so she at least knew which medication was which. And those experiences for me... It led me to want to just pragmatically make this better. And I think you're seeing a lot of that with general medicine, which is it's easy to dismiss as like, this is a convenience play. But if you look at happened with our customers at PillPack and I think what we're seeing at General Medicine is that convenience matters a ton if you want to help people be healthy. Arguably, it is one of the most important things to solve for in a world that this category has [10:00] so much friction in the way of doing things. And we're hearing from our customers at General Medicine that 80% of them have put off care for some amount of time, 30% have put off care for six months because it's so inconvenient to access. And so this has been the through line of our work is it's actually really important to try to make this stuff easier for customers. Yeah, I think I sort of see it in how your product and store is built, that you have a tremendous connection to the real world. You have a tremendous connection to what people are going to have to do. It's almost like the virtual store has like a... ramp out the back that lets you go into the into the other part of the store that's actually connected to the big bad world where there's physical care and pharmacies and someone who might need support physically, and you're not afraid to build that ramp. And I think that's sort of – you're understated about it, but it's hard to find in health tech. Well, I think if you really zoom out, it shows like the – Healthcare... is 10 to 15 years delayed from some of these other categories, right? And so, you know, in many ways, the teens in healthcare were like the e-commerce era in retail. And we're now clearly moving to a world where it's the marketplaces era and its connectivity into the real world. And if you just think about this in terms of these, like, evolutions of categories, that's been very obvious to us for a really long time. So I think you're really seeing... us bring that Same experience that you'd get from an Instacart or an Uber Eats, but in a far more complex category. Hopefully much faster this time. Hopefully much faster. Yeah. Let's talk about why it's so hard to make a store. You guys ended up at Amazon. Amazon kind of figured out how to make a store for a lot of other things. Yeah, they've been doing it for a while. You can shop for a lot of different products and do it pretty seamlessly. But they kind of need a needed some of the insights of PillPack to really launch the pharmacy store. I'm a proud Amazon pharmacy customer, very happy. Couldn't do it any other way. But why did these two pieces need to come together? Why was it so hard? *sniff* Yeah, I think there's a couple things there. The first is pretty non-intuitive. It's it kind of took us pounding our fist on the table, being like, the big opportunity in healthcare is just... being Amazon. It doesn't actually need to be different. And we were largely aligned with leadership on that strategy. I think the Practically difficult things about building Amazon Pharmacy are pretty unintuitive. The punchline is that we had to integrate into an ecosystem that was never intended to deal with PHI or be HIPAA compliant. And when we showed up, I was like, oh, we built this before. We're just building a UX on top of our infrastructure. It'll take us like six months and we'll launch this thing. It'll be great. And when you start to pull on that thread, if you want it to feel like it's cohesively part of the Amazon ecosystem, team that touches that CX. And so just to launch the first iteration of Pharmacy, which was like the kind of most bare bones version that we could. that we felt reflected what we were trying to achieve, it integrated with 40 different teams, none of which want to consume PHI, obviously. And so you had to do a bunch of technical work with each team to ensure that the consumer had a seamless experience. But we were being very thoughtful about how the technical integration worked. And this was across identity, payments, tax. Every little touchpoint required a bespoke approach. I think the good thing is that that integration work is done. So if they want to go launch three or four other healthcare things, that was a one and done exercise. But it was a ton of work just to get that initial launch out the door from a, from a technical standpoint. I think, It's easy to assume that the challenges were organizational. Like we didn't have alignment. We didn't have investment levels. We weren't on the same page. That was all actually great. Like that none of that was ever an issue. It was the sort of brass tacks of actually building and shipping the thing that was super hard. So what's going to be different about general medicine? What is already different about general medicine? Well, one is we don't have to integrate with a bunch of other infrastructure, so we can move a lot quicker. We're building de novo. I think if you go back to that PillPack Amazon experience, we'd built Popak from the ground up to be hit with compliance. It actually wasn't that difficult to do that right. It's much harder in the context of a broader ecosystem. I think the other thing is that, We're very committed to making it easy to access any type of care with any type of provider or service provider or pharmacy. And so we're very much leaning into this industry. connectivity into the offline ecosystem. And that's obviously very different than some of the work that we did in our past lives where we were the pharmacy, we built all the infrastructure. I think what we do over time realized was that customers want to be able to get care of wherever makes sense for them. And we want to be able to offer this same CX, but not be constrained to one person. [15:00] Ryder. It's almost like the... catalog of a general store Turns out it has to be general. Yep. And I think in many ways... You can't be a half. You can't have two aisles. Yeah, exactly. You can't just sell like sporting goods and be successful in this world these days. So, you know, I think... Think of the Art Mandalay import and export business. Import and export. Yeah. Like the... I think... You alluded to it, but the product catalog really was... one of the most important foundational inventions here. If you take pharmacy, There is a product catalog, right? It's a product. And they have NDCs, and we can define what the object is. that you're getting in healthcare that doesn't really exist. There's all sorts of similar codes that are designed for billing or other processes in healthcare. But no one has tried to translate the way the system thinks about things to the way the customer thinks about things. And so this product catalog mechanism really helps us do that. You're stepping... back into the CEO role at General Medicine for a second. The team is extraordinary. We are enormous fans of Elliott and Ashman and everything that they've built. And you were a co-founder and an investor at General Medicine really from the very beginning. And then you kind of had like your Travis Kalanick moment where – You're back, but you kind of also never left. Yeah, I kind of never left. You kind of never left. Yeah, but I'm more back. You're more back than I was. You're super back. What pulled you back in? I mean, it's super simple. It's more fun. It's so enjoyable to be able to be so deep in it and building the thing that the three of us, and to some degree the broader team, has wanted to exist for so long. I mean, we've been on this journey for 15 years, and as we keep pulling on these threads, the way we're going about this is the right way to help customers access healthcare. And so I was straddling, working on general medicine with the team and being a venture capitalist. And as the business continued to scale and see success and a number of the things that I was working on that were speculative started to come to fruition, I just got completely emotionally sucked back in. It's so fun to be back in the driver's seat and to be building the stuff that we have always wanted to build. Yeah. And the business is going to benefit from - Yeah. Elliot high-fived me yesterday for using the phrase GUMIX. I was pretty proud of myself, too. You're getting store-pilled. Not a concept that healthcare people think about, maybe often enough. As you mentioned, it's a core part of the design of the store. The store needs a big catalog. You can't run into a wall right when you need it most where the store says, sorry, we don't have that, and we also have no way to request it. So... How did you build a catalog that is so big? when so many other organizations and marketplaces have kind of carved out little pieces, but A, how do you define a product? in healthcare? And then how did you get so many products into the catalog? Yeah, I think to some degree it goes back to having built these primitives, right? So to have this breadth of selection, you have to be able to get people labs, obviously. You have to be able to prescribe medications. You have to be able to get them to a specialist. You need to have all these tools available to you to deal with more and more complex things. So to some degree, the ability to spin up a very broad catalog is contingent on being able to help you across all of these journeys. So that's step one. And then step two is turning these things into SKUs, right? And so what goes into that is a couple things. The first is that each SKU, so it might be shoulder pain, it might be an MRI, it might be a set of labs, like each one of those has a defined set of clinical criteria, where that's an appropriate choice for the patient. consumer. And that also has an associated CIQ to gather that information and to know whether it's appropriate for the consumer. So each SKU is going to have slightly different CIQs and even may have multiple CIQs depending on, you know, whether taking a beta blocker for anxiety or for blood pressure, right, is going to be a different set of clinical criteria. So that's, the sort of first half of it. And the second is we need to take all of these billing codes and other things on the back end and align that to the skew. So if you're getting colonoscopy A versus colonoscopy B, the consumer might think of those things as similar, but under the hood, they're going to get billed for very differently. They're different purposes. And so you got to be able to parse that and have this crosswalk from what's going to get billed to what the consumer thinks they're buying. And then you also have this overlay of you need to allow providers to say like, this is a skew that we can service. If you think about the magic of The benefit is 2,000 sellers can all sell the same thing. What that does for customers is it drives the price down and it creates competition. And so we need the same context here where we know what providers are appropriate for each one of these customers. individual skews. [20:00] who's in the same DERM that's going to be, wants to do a mole check, may not want to do a bunch of other stuff. And so you got to be really smart about getting the customer to the right seller, the right provider. And so this... Think of the product catalog really as this translation layer between consumer wants something And there's a certain set of people that can provide it and they're going to bill for it in a very specific way. And we need to be able to do that translation for the end customer. So it's a classic two-sided marketplace. Yep. The SKUs define the catalog sort of in the middle, and then there's a bunch of supply and then a bunch of demand. In marketplaces, you often think about, okay, which side has to get going first? Do I need to first confirm that there's demand before supply will sign up? How expensive is it to sign supply up? Do you have to go door to door and ask every cardiologist in America if they want to play ball with you? What's it like to create supply that is willing to service that stew mix? And is it harder or easier than your typical marketplace? Yeah, I think we've made the bet that you need to get consumers first, and you need to be able to offer them a broad set of things first. And from a strategic standpoint, approaching this in the ways that the most successful hybrid first-party, third-party marketplaces have come to fruition, right? And so rather than having to go pound the pavement to sign up a bunch of providers, we made the decision early on that we will build our own provider group so that we can serve a broad set of needs. So we both have our own medical group that is effectively doing the work that a GP would do in the normal circumstances. with that provider group. So to your earlier point, we can get you e-consult with a specialist same day, which even if you ultimately need to get seen in person with a specialist is going to fast track that process. And then I think the other thing we did that Um... is novel in the category, is we can get you to any provider for any service, not contingent on them participating in the marketplace. And that comes back to the fact that we behave like a GP. And in the healthcare marketplace, the GP can refer you to whatever clinician is appropriate for you. And so we are not constraining who we can help you access based on deep integrations or financial relationships or any other constraint. And we have the assumption that over time, we'll go deeper deeper from an integration standpoint with health systems and provider groups across the country. But we're really leaning into offering the full offering to the customer first. And over time, building out the breadth of deep integrations and integrated providers. And then we're, you know, in other categories that are not provider categories, take pharmacy. We've built a lot of stuff in pharmacy, and we launched an experience a couple months ago where you can check out on General Medicine at 50,000 plus effectively any pharmacy in the country. And so that does feel to the consumer like a deep integration that went into that one. And so it's this combination of where we can be really clever to have deep integrations nationwide. We're really clever about that stuff. Where there is just a lot of work to do and slog on integrations with folks like health systems and provider groups. We want to get you to whoever's appropriate for you. And we're assuming that over time we will integrate more and more of that selection. So you had a hack with your own medical group. uniquely designed to be a multi-specialty group. But I do think it's worth underscoring that healthcare is a little different in this way than like the shampoo supplier who has to decide to ship shampoo to Costco. Like the healthcare specialist receives referrals today inbound anyway. And you actually offer them a zero CAC service customer acquisition channel, which has always been true in healthcare actually, because Because they're not demand constraints. The referral is the ultimate kind of zero-cac way to not only earn a customer, but to see someone who really needs your care. And so you have this kind of unique... ability to draw supply on, how is that going in terms of formalizing any partnerships on the, on the, on the supply side? Or is that not even needed? Are you, are you guys just going to call the cardiologist and say, hey, we got a referral for you. We're submitting it. It's yes and. So we don't want to constrain who we can get you access to so that, again, we're not going to constrain the supply based on these integrations. But we think it is a better consumer experience if, like in labs, you can book the thing like an open table experience versus us doing the work to help you get the appointment. And so these integrations for us are primarily motivated around how do we just help you get even better access, faster access, seamless access, and we're pursuing that. as fast as we can, both on the health system side. So helping figure out who's appropriate for you, what the criteria is, so that you can go straight to a procedure versus having to do a workup first. [25:00] And then on the telemedicine side where there's opportunities to partner with other telemedicine providers that want access to customers and we want to make it easier for customers to access them. And I think similarly in telemedicine, there's lots of different provider groups that have, you know, better copay structure with certain payers. And we just want all those options to be available to the end customer. So we are definitely working hard to integrate more and more providers at the same time, ensuring that the customer experience is not constrained to that provider set. It might be the most... General, most comprehensive store in healthcare. Broadest store in healthcare. Let's talk about the role that AI is playing today. In your product. Yep. So, you know, We've noticed in specialty clinics all over the country, I had an oncologist tell me, hey, patients with diagnosed with newly, you know, newly diagnosed colon cancer are coming in, and they're asking when they start Fulfox, right? When do they start the chemotherapy regimen that they already know is very likely to be indicated in a case? That's a different world, sort of a similar analogy to the primary care visit split. You know, you don't spend a visit necessarily explaining... step zero in a care journey. You get to jump to a more interesting, richer conversation about what's going to happen to me, what was my prognosis, what are my options. How do you facilitate that with AI inside the general medicine product? Yeah, I mean, I think there's two sides to that dynamic. The first is the one you're mentioning and we've talked about, which is consumers are already doing this, right? They're already going deep on what their options are and trying to figure out what the best path is. On general medicine specifically, they can do that in the context of their total medical record. So our GPT, the consumer can go back and forth and it has the full context of their history. And I think importantly different than the kind of consumer-directed version of that experience where the record density is much lower. So there's just a lot that has a very similar set of tooling. And so we're taking that full record, synthesizing it into the GPT for the provider, working really hard to summarize exactly where that customer is on their journey so that when they show up to the visit, they're ready to have this deeper dialogue. And we want that to be this full loop so the customer can keep bopping back and forth with the GPT after the visit. All that gets synthesized again for the provider next time. And so it really is closing this loop between the two participants. with AI on the provider side is because we have all these primitives, the GPT can recommend a set of next steps for the customer, whether it's specific to what they came for. So maybe they just needed a simple medication, but they're 55 and they haven't had a mammography or they're 50 and they've never had a colonoscopy. The GPT is going to recommend those next steps for the provider and compared to a normal patient. experience where it's like, oh by the way, you should go do this. Or maybe at best, like let me write you a referral to one place to get that mammography. Each one of those next steps is a transactable skew for the consumer. So you know, after a visit, you'll get the script that you came for, but you'll also get recommended next steps that you can immediately go execute. In the same store. In the same store. It's all in one place. And so I think that is the magic of, it's really this collaborative back and forth between the customer, the provider, and then a system that sits underneath that, that takes all of those things and makes it just really easy to go action. One of the things I love about this is... Like the provider... On your supply side, is like constantly getting better at their job. Like they're like, I get to do the fun stuff. Yeah. I have more tools. Totally. You take the referral, like, you know, or you take the, you take a prescription. Like if, One GLP-1 is a preferred drug on the formulary. The customer just wants the one that they can get covered. Providers have gotten sucked into this world of having to go figure all that out, which one's covered, which one's uncovered, which one costs what. The customer cares a lot more. We give them the tools that can help make those decisions, and providers can really focus on clinical guidance and advice. Or give them the one that's not covered, and let's talk about the difference. That whole conversation, they're just so super empowered. that's enabled by all the tools and technology that you give. It seems like you also have a unique opportunity, which marketplaces tend to have, to actually see how well they're doing, So some marketplaces rate based on... demand side ratings. And so you could imagine that providers are stack ranked in some way based on how much they are adopting those tools, and how much they are trying to get better and better. You could also imagine kind of finer... finer-grained Um, [30:00] description of the supply side in such a way that's super matched to what the patient actually needs. Everybody's had the experience where they bop around to three or four doctors before getting to the right one. Like what are, what's the role of general medicine in curating, what's the store and actually recommending products in the store. Yeah, I think there's multiple vectors of what at any given moment a customer might care about, right? So there's convenience and speed. Sometimes that's the most important. Wait, who can you get to see now? Yeah, even take the biopsy example. Speed is the most important thing. And there's a whole spectrum of providers that can offer the biopsy that don't have to be connected to where you ultimately might want to get treatment. And so in that scenario, the customer should be optimizing rare form of cancer, you probably want to get to the guy. So it's a whole different set of criteria that matter to you in that moment. And so I think being really smart in this shopping experience to help elucidate which scenarios are we dealing with. And again, that isn't just the consumer on their own, like, I hope I can figure this out. We will play the role of helping you figure that out as clinicians. But at the end of the day, we want all of those different frameworks and vectors and the way consumers care about it to be just easy for them to make this decision in collaboration with a clinician. Let's talk about the payment side of it, because a core part of your infrastructure is extraordinary price transparency with full knowledge of a patient's coverage. So you're not guessing. You figured it out. It was magical for me when I... gave you my driver's license and you figured out everything about what's covered for me. How do you... First of all, how do you do that? Yeah. And this is another place where you need LLMs to make this possible. You couldn't have built this five years ago. And so under the hood, what we're doing is we can crosswalk from your demographics to your insurance information. And it's not just... member ID and, you know, bin and PCN. Like, it's not just the things on your card, though. We obviously get that. But we're going all the way to your full COB and using LLMs to scrub your COB and turn that into structured data so you understand what you're eligible for, what the copay structure is, what the out-of-pocket structure is. We're then pulling down your current personalized deductible status and out-of-pocket status and... consuming the open pricing files that exist so that we know what the provider is going to get reimbursed for for any given procedure. And then we mung that all together. And the way it manifests for the customer is it's $30 to go do this office visit. And I think to the customer, it's like, oh. It just works like everything else. Under the hood, you've got this kind of triple Lindy happening to make that possible. But really for the customer, it's just like, let's make this feel like it's just sort of simple and magic. And I do think... Even to your point, just being able to snap straight to, like, I got your benefit information. You didn't have to, like, take a picture of it. You didn't have to do anything. I didn't take a picture of my insurance card. Like, we just, we got it. Yep, you got it. And in many ways, I think that's the... the vibe that we want for our customers is like, we just got this. We figured it out for you. And that's not just constrained to pricing. It's really like we want that role across the board. Some people argue that price transparency doesn't matter in healthcare because like I don't really have a choice if I need healthcare. And I'm kind of like, maybe I'm price and you know, maybe my demand for healthcare is actually price inelastic at some level, because I really need it. Yeah. Like, what's your what's your response to that? And I know you had A lot of direct. experience with this elasticity curve at PillPack. Yeah, I think that's nuts, generally speaking. Obviously, inpatient, different thing. But the vast majority of medicine, I think that's kind of a crazy point of view to have. And a good anecdote is that when we built Pharmacy the last go-around, all we did was show cash insurance and carton checkout. We didn't spend a ton of energy trying to optimize Let's just show you like a low cash price next to your insurance price. And we saw nearly half of units go from on benefit to cash. And if you compare that to the status quo in normal retail pharmacies, 5, 6, 7% that are being paid with cash or GoodRx or other kind of non-reimbursed mechanisms, and we were 10x the share shift to cash. And I think the The interesting thing about that, it was 25% of the time it was actually cheaper. So kind of obvious that... But not all the time. But not all the time. So 25% of the time it was cheaper and... by and large, customers are going to pick the cheaper option. So you're going to win those sales most of the time, win those sales. But the other 25%, it was more expensive than the cash price. But all of a sudden, all the sad past that insurance creates for you are now just gone, right? So maybe it was 17 bucks instead of 12 or 21 instead of 18 or 15 instead of five. But the transaction's done. But it's just done. It's like, [35:00] You don't have to wait four days to refill it because you didn't perfectly align the timing of the refills or you didn't have a prior auth. You had to go chase down and deal with that whole thing. It just was like, oh, I just buy the thing and I'm willing to spend six dollars more to get it today. Am I going to want to deal with this complexity? And obviously the shape of that across different categories is going to be different. But I do think it's just a really interesting anecdote that by doing something like what feels like it should be table stakes and relatively simple had such a dramatic impact on. on customer behavior. And I think the other trope that people have here is that the customer is not the customer, right? The customer is the employer, it's the federal government, it's the ultimate payer, and therefore, like, yeah. They're insured. So, like, they're not the customer, so they're never going to, like, be the one that makes these decisions. I think that's crazy. They're at the call face of, like, what they're going to do for their health. You take this pharmacy example, it's like, if you believe that, in its most simplistic way, are they the customer the half of the time they chose cash and they're not the customer now that they chose? It's illogical. And so I do think like, I think showing customers these options will really impact their behavior. I guess the last point here is that Pairs have spent 20 years trying to be really sophisticated about how to price these things in a way that drives the right customer behavior. They want to get you to the lowest cost, highest quality provider they can, but there's been very little energy in ever communicating any of that. to the customer. And so in many ways what we're trying to do is say, you've done a lot of work here, and a lot of it's actually pretty good work, but it is completely lost on the end user, like What the hell is going on? Right. And so I think we view our role as like actually helping people understand these things. And we think that will like eliminate both a lot of that friction there, but also drive the behaviors that I think we all think are the most effective. 12 versus 17 seems low stakes, but wait till you try to schedule an MRI or an echocardiogram. Yeah, like I got an MRI for my knee, which I tore my ACL back in the spring. I wonder what you were doing. I was skiing with Elliot in Alaska. It's really shocking. That's when I was a venture capitalist, so I was allowed to ski then. You're still not skiing, you just gotta take me with you. And it was $1,700 for the MRI in Park City at Intermountain with the same insurance, and it was $600 in Salt Lake at Intermountain with the same insurance. And this stuff is completely invisible. Yes, even if you were freaked out about interoperability. Totally. You know? near identical in this huge variation. And Every day, customers are making all these decisions without any of that information. It's just, that has to change. What do you think about your partnership with organizations. who like you said, have tried to solve these problems for decades. Yep. Um, Health systems don't really love that that's happening either. neither do the payers. So it seems like there's an obvious way for you to cut in and say, hey, we'll be the right interface, your member, your patient, your consumer, whatever you want to call them. And we're just treating them as people who want to make smart decisions in a store. Is there an example you can share of how you would fit in to a large incumbent organization like that? Yeah, I think we view... what we're best at is building this customer experience for the end customer. It is this translation layer from traditional healthcare to the way we think customers want to consume this care. And I also think we think that this idea of there's going to be a singular front door is nuts. People are going to access healthcare in all sorts of places, in all sorts of ways. Wherever they get it, it'll be in your record. Yeah, exactly. Yeah. participants in the ecosystem to offer their customers, their members the same experience. And so that could take the shape of a large health system, which has incredible clinical care, but very complex to get access to that care. I think you see us show up in that scenario as just a really amazing way for customers to better understand how to get into that care. And we're working through experiences like that. It could be a large retailer that wants to offer their own that wants these benefits to be more understandable to their members. And so I think we're, We're exploring a number of similar partnerships where others in the ecosystem can build their store and offer this experience to their customers. Like a white label catalog. Yep. You talked about a bunch of transactions that are going to happen in the general medicine store. Yep. How does the storekeeper, general medicine, make money and be sustainable? Yeah, so the best... proxy for the way that we've built this business is a successful first-party and third-party marketplace. Obviously, there's a lot of nuance and complexity in healthcare, so it's not as [40:00] by category. I think the other thing that we're seeing is similar to consumers' willingness to pay for convenience on DoorDash or on Uber Eats. If you can facilitate a booking with a specialist and provide the referral, they're happy to pay for that convenience. We are creating a lot of value for that customer. And so I think that is... not something that has proliferated healthcare yet, but we are seeing that customers are perfectly happy to pay for simplicity, access, convenience. And so again, there's good analogies for how the business writ large ultimately makes money, but that's probably the most novel thing in healthcare that we haven't really seen otherwise. And that'll be especially true if you're saving the patient a couple hundred bucks scheduling something because you created transparency along the way. Totally. It's crazy how much, probably most people have this experience. If you're looking for healthcare, I have a thing to do. I need to go schedule this appointment. I need to go make it happen. Your vision where it just gets done. Can't wait to do it together. Thanks, TJ. Thanks for listening to this episode of the A16Z podcast. If you liked this episode, be sure to like, comment, subscribe, leave us a rating or review and share it with your friends and family. For more episodes, go to YouTube, Apple Podcasts and Spotify. Follow us on X at A16Z and subscribe to our Substack at a16z.substack.com. Thanks again for listening and I'll see you in the next episode. only, should not be taken as legal business, tax, or investment advice, or be used to evaluate any investment or security, and is not directed at any investors or potential investors in any A16Z fund. Please note that A16Z and its affiliates may also maintain investments in the companies discussed in this podcast. For more details, including a link to our investments, please see a16z.com forward slash disclosures. Thank you. you Thank you.