在a16z最近一期于2026年9月16日录制的播客中，乔·施密特（Joe Schmidt）和亚历克斯·兰佩尔（Alex Rampell）采访了莱特菲尔德（Lightfield）的首席执行官基思·帕里斯（Keith Paris）。莱特菲尔德是一家最近完成由a16z领投的4700万美元A轮融资的公司。莱特菲尔德正在构建一个“商业世界模型”，将客户互动转化为可供AI代理操作的记录。 基思分享了他从之前的创业公司托姆（Tome）到莱特菲尔德的“非常规旅程”。托姆最初是一个面向消费者的产品，专注于使用早期AI（GPT-3）生成演示文稿，实现了“爆炸式增长”，每月拥有“两百万用户”。然而，尽管规模庞大，基思和他的联合创始人发现他们并不“热爱这个产品”。他们意识到托姆无法创建专业用途（例如，投资银行、咨询）所需的高质量、高鉴别度的演示文稿，因为AI缺乏关于演示者、受众及其关系的足够背景信息。这导致了他们做出艰难的转型决定，因为他们看不到明确的路径能将其打造为目标高端用户的不可或缺工具。 转型后，他们寻求B2B用例，特别是在销售和市场营销领域。通过与公司的试点项目，他们发现企业不仅仅是寻找AI来“完成工作”（如潜在客户评分或发送电子邮件），而是要“理解所有这些不同系统中的所有数据”。真正的问题在于现有CRM中数据的“不完整”和“冲突”性质，使得“重组数据”成为最关键的任务。这一洞察让他们意识到，如果他们能“完全建模你的业务和客户现实”，其余的问题就会迎刃而解。 莱特菲尔德的架构是为此目的有意设计的。受Facebook时间轴的启发，他们的五位创始成员中有三位将“活动日志”作为核心原语，按时间顺序记录每一次互动——电子邮件、电话、会议、文档、产品使用、支付。这种“规范的关系日志”允许系统推断因果关系并触发传统的CRM更新。他们采用了“无模式设置”，用户连接数据源（电子邮件、数据仓库），莱特菲尔德会自动构建关系，使其设置起来感觉像消费产品一样。这种“智能优先于模式”的方法使其与僵化、老旧的CRM区别开来。 莱特菲尔德影响力的一个引人注目的例子是“动力”（Power）公司，该公司利用莱特菲尔德来建模制药和医疗保健领域中复杂的企业对企业和企业对消费者的互动。“动力”公司汇集患病个体，并将他们与前沿治疗和临床试验联系起来。通过抓取FDA数据和clinicaltrials.gov上的信息，莱特菲尔德帮助他们匹配患者，甚至能帮助阿尔茨海默病患者“在几天内”找到治疗方案。 基思讨论了他们驾驭CRM“红海市场”的策略，对比了“绿地市场”（新公司）和“棕地市场”（使用现有厂商的成熟公司）。最初，莱特菲尔德瞄准了绿地初创企业，甚至提供“负定价”（免费办公空间）来吸引早期用户。这使他们能够与快速增长的公司共同发展，其中一些公司在使用莱特菲尔德的同时，销售人员从零增长到“上百人”。这段经历让他们认识到，“棕地市场的切入点在于更好地理解你的公司”以进行战略指导，而不仅仅是自动化任务。 针对传统销售副总裁们“习惯并受训于”像Salesforce这样的老式CRM的挑战，莱特菲尔德允许整个公司免费使用其产品。这创造了“公司网络效应”，工程师、财务和客户成功团队都依赖莱特菲尔德，使得新的销售副总裁更难要求更换系统。 莱特菲尔德的定价从纯按席位或纯按使用量计费模式的失败尝试中演变而来。他们最终确定了“平台费加核心CRM席位费”的模式，涵盖基本功能。对于渠道生成、工作流自动化和智能/预测（提供明确投资回报或“阿尔法”）等增值服务，他们则按使用量计费。 关于公司文化，莱特菲尔德强调速度。拥有40名员工的他们秉持“人人都是多面手”的理念，即“人人负责产品，人人负责客户成功”。他们进行每日站会和持续规划，秉持“项目启动门槛低，但交付门槛高”的原则。基思将其归因于AI时代，在这个时代，工具让任何人都能迅速了解客户需求、设计系统或进行任务管理。 基思最大的担忧是速度，确保他们足够快地进行建设，以防止客户产生回归“旧世界”的愿望。他很期待莱特菲尔德能成为战略情景规划的“水晶球”，帮助公司回答诸如“我应该雇佣多少销售代表”或“我下一步应该开发什么产品”等关键问题。 对于考虑转型的创始人，基思的建议是“戴上眼罩”。他强调“当你处于转型期时，周围几乎所有的噪音都不重要”，敦促创始人“发现痛点”，“受启发去构建解决该痛点的产品或服务”，并“狂热地专注于你的客户”。

In a recent a16z podcast recorded on 2026-09-16, Joe Schmidt and Alex Rampell interviewed Keith Paris, CEO of Lightfield, a company that recently raised a $47 million Series A led by a16z. Lightfield is building a "business world model" that transforms customer interactions into actionable records for AI agents. Keith shared his "very atypical journey" from his previous venture, Tome, to Lightfield. Tome began as a consumer product focused on presentation generation using early AI (GPT-3), achieving "explosive growth" with "two million users a month." However, despite its scale, Keith and his co-founders found they didn't "love the product." They realized Tome couldn't create the high-quality, discerning presentations needed for professional use (e.g., investment banking, consulting) because the AI lacked sufficient context about the presenter, audience, and their relationship. This led to the difficult decision to pivot, as they saw no clear path to making it an indispensable tool for their target high-end users. After the pivot, they sought B2B use cases, particularly in sales and marketing. Through pilot programs with companies, they discovered that businesses weren't just looking for AI to "do the work" (like lead scoring or email sending), but rather to "make sense of all of the data across all of these disparate systems." The real problem was the "incomplete" and "conflicting" nature of data in existing CRMs, making "reorganizing it" the most crucial task. This insight led to the realization that if they could "completely model your business and your customer reality," the rest would follow. Lightfield's architecture was intentionally designed for this. Inspired by Facebook's timeline, three out of their five founding members built an "activity log" as the core primitive, chronologically recording every interaction—emails, calls, meetings, documents, product usage, payments. This "canonical log of relationship" allows the system to infer causality and trigger traditional CRM updates. They embraced a "schemaless setup" where users connect data sources (emails, data warehouses) and Lightfield automatically assembles relationships, making it feel like a consumer product to set up. This "intelligence is greater than schema" approach differentiates it from rigid, older CRMs. A compelling example of Lightfield's impact is "Power," a company that uses Lightfield to model complex business-to-business and business-to-consumer interactions in the pharmaceutical and healthcare space. Power aggregates individuals with illnesses and connects them to frontier treatments and clinical trials. By scraping FDA data and clinicaltrials.gov, Lightfield helps them match patients, even aiding someone with Alzheimer's to find treatment "within days." Keith discussed their strategy for navigating the "red ocean space" of CRM, contrasting "greenfield" (new companies) with "brownfield" (established companies using incumbents). Initially, Lightfield targeted greenfield startups, even offering "negative pricing" (free office space) to attract early adopters. This allowed them to evolve with fast-growing companies, some going from zero to "hundred reps" while using Lightfield. This experience taught them that the "wedge in brownfield has to do with like better understanding your company" for strategic steering, rather than just automating tasks. Addressing the challenge of traditional VPs of Sales who are "acclimated and trained" to older CRMs like Salesforce, Lightfield allows entire companies to use the product for free. This creates "company network effects," where engineers, finance, and customer success teams all rely on Lightfield, making it harder for a new VP of Sales to demand a switch. Lightfield's pricing evolved from failed attempts at pure seat-based or pure consumption-based models. They settled on a "platform fee plus seat for core CRM," covering essential functions. For value-added services like pipeline generation, workflow automation, and intelligence/forecasting (which offer clear ROI or "alpha"), they charge on a consumption basis. Regarding company culture, Lightfield emphasizes velocity. With 40 employees, they operate with a "everyone is a generalist" philosophy, where "everyone owns product and everyone owns customer success." They conduct daily stand-ups and continuous planning, with a "low bar to start a project" but a "high bar to ship." Keith attributes this to the AI era, where tools enable anyone to quickly ramp up on customer needs, design systems, or task management. Keith's biggest worry is speed, ensuring they build fast enough to prevent customers from feeling the desire to return to "the old world." He's excited for Lightfield to become a "crystal ball" for strategic scenario planning, helping companies answer critical questions like "how many reps should I hire" or "what products should I build next." For founders considering a pivot, Keith's advice is to "put the blinders on." He stresses that "almost none of the noise around you matters when you're in a pivot," urging founders to "find pain," "be inspired to build a product or service that solves that pain," and be "maniacally focused on your customers."