We're sunsetting PodQuest on 2025-07-28. Thank you for your support!

Balancing AI Expertise and Industry Acumen in Vertical Applications

2024/9/13

AI + a16z

AI Deep Dive AI Insights AI Chapters Transcript

People

Nikhil Buduma

Topics

Nikhil Buduma: Ambience公司是早期采用Transformer架构的团队之一，在将Transformer模型投入生产的过程中积累了宝贵的经验。通过与OpenAI和Anthropic团队的密切合作，我们对AI领域的进展有了深入了解，这帮助我们解决了之前难以解决的临床问题。从医学研究到机器学习，再到医疗保健创业，我的职业生涯经历了转变，并与Jeff Dean和Sam Altman等业界领袖建立了联系。我们相信，强大的通用人工智能模型将对社会产生巨大影响，医疗保健将是这些技术最有意义的应用领域之一。与导师的合作，巩固了我们对AI在医疗保健领域应用的信心，并教会了我们敢于冒险的精神。在Remedy Health，我们学习了医疗保健行业的复杂性，包括医学科学、医疗系统和医疗经济学。我们开发了一系列技术，包括聊天机器人、AI辅助电话筛选工具和基于Transformer的预测模型，以改善医疗服务。IBM Watson的失败案例表明，仅仅关注算法优化而不考虑数据集成和工作流程，会导致产品难以被采用。Remedy Health的经验使我们深刻理解了临床医生的需求，即使这些需求并非总是最引人注目的。Transformer架构的出现以及对AI领域进展的深入了解，促使我们创建了Ambience公司，专注于将先进的AI模型应用于医疗保健领域。Ambience公司致力于利用AI来解决医疗保健领域的人员短缺和行政负担问题，首先从医疗文档自动化入手。我们花了数年时间才达到医疗文档自动化产品所需的质量水平，以确保临床医生的采用。我们密切关注AI领域的最新进展，并根据自身需求和未来发展趋势进行研发投资。垂直型AI公司需要选择合适的用例、构建合适的模型、集成数据、设计产品、并注重交付能力和变更管理。在选择基础模型时，应避免与大型基础模型厂商直接竞争，因为他们拥有更强的资源和更广泛的市场覆盖率。在医疗保健领域，现成的基础模型在性能上存在局限性，因此我们需要对模型进行大量投资。医疗保健领域的多样性和低利润率，要求我们在模型堆栈和企业价值创造方面进行大量投资。对于大多数组织来说，构建基础模型可能不是最佳选择，而垂直整合的应用层投资更有价值。构建成功的垂直AI公司需要具备行业专业知识、机器学习能力、产品设计能力、交付能力和变更管理能力。 Derrick Harris: 积极参与讨论，引导话题，并就AI技术在医疗保健领域的应用提出问题。

Deep Dive

Key Insights

Why is industry expertise as important as technical AI knowledge for building vertical applications?

Industry expertise helps map out the most valuable use cases and understand how they synergize, which is crucial for creating compounding value. Technical AI knowledge alone is insufficient; integration with enterprise data, workflow design, and change management are equally critical for adoption and success.

What was Nikhil Buduma's early experience with AI and healthcare?

Nikhil co-founded a company in 2013 using convolutional neural networks for low-cost malaria detection. He authored a textbook on deep learning and worked with mentors like Jeff Dean and the early OpenAI team, which shaped his conviction about AI's potential in healthcare.

How did Nikhil's personal health experiences influence his career path?

Nikhil grew up with chronic heart conditions and saw his parents struggle with the U.S. healthcare system, which solidified his commitment to working in healthcare and improving the system through technology.

What challenges did Nikhil face while building Remedy Health?

Remedy Health faced technical limitations, market readiness issues (e.g., consumer reluctance to adopt virtual care before the pandemic), and the complexity of building a medical practice from scratch, including hiring doctors and navigating insurance systems.

Why did Nikhil and his team shift from Remedy Health to Ambience?

The shift was driven by the rapid advancements in AI, particularly the transformer architecture, which made previously intractable problems in healthcare seem solvable. Ambience focused on leveraging these advancements to create a better-integrated platform for healthcare institutions.

What percentage of a clinician's day is spent on direct patient care?

Only about 25-27% of a clinician's day is spent on direct patient care, with the rest consumed by administrative tasks like documentation, coding, and prior authorization.

How does Ambience approach integrating AI into healthcare workflows?

Ambience focuses on solving specific pain points like medical documentation, which clinicians spend a quarter of their day on. They aim to automate these tasks to free up more time for direct patient care, ensuring the AI solution is robust and integrates seamlessly with existing workflows.

Why do off-the-shelf AI models struggle in healthcare applications?

Off-the-shelf models often hit performance ceilings quickly due to the complexity and esoteric nature of healthcare knowledge, which is typically passed down through apprenticeship rather than public domain information. This makes fine-tuning and domain-specific models essential.

What advice does Nikhil have for building teams for vertical AI applications?

Founders should deeply understand their industry, map out valuable use cases, and ensure their team has the right mix of ML expertise and industry knowledge. Collaboration between product managers, engineers, and domain experts is crucial for building effective solutions.

How does Ambience handle the balance between automation and clinician autonomy?

Ambience recognizes that not all clinicians will adopt new technology immediately. They focus on creating nuclei of success within institutions, where early adopters champion the technology, gradually expanding its use while addressing any product limitations promptly.

Chapters

Nikhil Buduma's childhood health issues led him to pursue healthcare. His research combined biology, medicine, and machine learning, culminating in a book on deep learning and co-founding a healthcare startup. Early exposure to AI advancements shaped his conviction in its potential for healthcare.

Early career focused on healthcare due to personal experience with chronic illness
Combined research in biology, medicine, and machine learning
Co-founded a healthcare startup
Early adoption of transformer architecture

Shownotes Transcript

In this episode of AI + a16z, Ambience) cofounder and chief scientist Nikhil Buduma joins Derrick Harris to discuss the nuances of using AI models to build vertical applications (including in his space, health care), and why industry acumen is at least as important as technical expertise. Nikhil also shares his experience of having a first-row seat to key advances in AI — including the transformer architecture — which not only allowed his company to be an early adopter, but also gave him insight into the types of problems that AI could solve in the future.

Here's an excerpt of Nikhil explaining the importance of understanding your buyer:

"If you believe that the most valuable companies are going to fall out of some level of vertical integration between the app layer and the model layer, [that] this next generation of incredibly valuable companies is going to be built by founders who've spent years just obsessively becoming experts in an industry, I would recommend that someone actually know how to map out the most valuable use cases and have a clear story for how those use cases have synergistic, compounding value when you solve those problems increasingly in concert together.

"I think the founding team is going to have to have the right ML chops to actually build out the right live learning loops, build out the ML ops loops to measure and to close the gap on model quality for those use cases. [But] the model is actually just one part of solving the problem.

"You actually need to be thoughtful about the product, the design, the delivery competencies to make sure that what you build is integrated with the right sources of the enterprise data that fits into the right workflows in the right way. And you're going to have to invest heavily in the change management to make sure that customers realize the full value of what they're buying from you. That's all actually way more important than people realize."

Learn more:

Fundamentals of Deep Learning)

Follow everyone on X:

Nikhil Buduma)

Derrick Harris)

Check out everything a16z is doing with artificial intelligence here), including articles, projects, and more podcasts.

Balancing AI Expertise and Industry Acumen in Vertical Applications 42:59 Share