How AI Is Turning Figma’s Finance Team Into Builders

在本期 A16Z 播客中，主持人克米·安布尔（Cmi Amble）和伊万·马卡罗夫（Ivan Makarov）邀请了 Figma 的首席财务官普拉维尔·马尔瓦尼（Pravir Malwani），共同探讨自动化和人工智能如何改变财务职能，将首席财务官的角色从“交通警察”转变为运营领导者和业务赋能者。 马尔瓦尼指出，在他的职业生涯早期，在当前这波人工智能工具出现之前，自动化意味着编写冗长的 SQL 查询、构建数据表，以及从 Stripe 和 Salesforce 等各种系统中手动拼凑信息。如今，借助现代人工智能工具和大语言模型，财务专业人员能够前所未有地获取实时数据，从而能够按小时或按天分析异常情况，而不必等待月度或季度周期。 在为团队招聘时，马尔瓦尼寻找具备强烈求知欲、数据流畅度以及提出扎实问题并有效构建思维能力的候选人。他强调，虽然人工智能工具可以快速生成模型输出和文档，但候选人必须仍然是严谨的思考者，不能将思考外包给机器。在面试过程中，案例研究部分依然神圣不可侵犯，用于测试应聘者是否真正理解他们所构建模型背后的结构、敏感性和直觉。 马尔瓦尼分享了 Figma 团队如何利用自动化和人工智能的几个例子。信贷和催收团队简化了发票跟踪、现金催收和 NetSuite 对账，释放了构建自动化信贷检查工作流程的产能。同时，企业发展团队在尽职调查期间使用人工智能工具——例如通过氛围编程（vibe-coding）产品来测试技术护城河的深度——并针对高管的模拟个性运行投资备忘录，以完善论点并缩短会议周期。其他团队通过在 Navan、Zip 和 Ramp 等不同的数据源之间执行模糊匹配，实现了差旅和娱乐（T&E）政策审查的自动化。 针对人工智能生成的“垃圾内容”或低质量文档的风险，马尔瓦尼强调，任何发布给他人消费的内容都需要时间和精力，这意味着员工仍然必须进行审查、总结并保持高标准。然而，如果深思熟虑地使用，人工智能可以提供巨大的速度和迭代优势。 展望未来，马尔瓦尼认为，如果业务复杂性保持不变，由于效率的提高，未来组织将不需要那么多的员工人数。他将现代技术的获取以及构建内部代理和工具的能力视为一个强大的招聘载体，能够吸引希望学习和成长的顶尖人才。归根结底，马尔瓦尼强调，伟大的财务领导者必须充当业务运营合作伙伴，深入理解业务，推动执行，并利用现代工具简化复杂性并引领公司向前发展。

In this episode of the A16Z Podcast, hosts Cmi Amble and Ivan Makarov are joined by Pravir Malwani, CFO at Figma, to discuss how automation and AI are transforming the finance function, shifting the role of the CFO from a "traffic cop" to an operational leader and business enabler. Malwani notes that early in his career, before the current wave of AI tools, automation meant writing long SQL queries, building data tables, and manually stitching information together from various systems like Stripe and Salesforce. Today, with modern AI tools and LLMs, finance professionals have unprecedented access to real-time data, enabling them to analyze anomalies on an hourly or daily rhythm rather than waiting for monthly or quarterly cycles. When hiring for his team, Malwani looks for candidates who possess strong intellectual curiosity, fluency with data, and the ability to ask solid questions and structure their thoughts effectively. He highlights that while AI tools can generate model outputs and documents quickly, candidates must still be rigorous thinkers who do not outsource their thinking to the machine. During the interview process, the case study portion remains sacred to test whether applicants truly understand the structure, sensitivities, and intuition behind the models they build. Malwani shares several examples of how his team at Figma leverages automation and AI. The credit and collections team has streamlined invoice tracking, cash collections, and NetSuite reconciliations, freeing up capacity to build automated credit-check workflows. Meanwhile, the corporate development team uses AI tools during diligence—such as vibe-coding products to test the depth of a technology's moat—and runs investment memos against simulated personalities of executive leaders to refine arguments and reduce meeting cycle times. Other teams have automated T&E policy reviews by performing fuzzy matches across disparate data sources like Navan, Zip, and Ramp. Addressing the risk of AI-generated "slop" or low-quality documents, Malwani emphasizes that anything put out for others to consume requires time and effort, meaning employees must still review, summarize, and maintain high standards. However, when used thoughtfully, AI provides a major speed and iteration advantage. Looking ahead, Malwani believes that if business complexity were to stay flat, organizations wouldn't need the same headcount in the future due to efficiency gains. He views modern technology access and the ability to build internal agents and tools as a powerful recruiting vehicle, attracting top talent who want to learn and grow. Ultimately, Malwani stresses that a great finance leader must act as an operational business partner who understands the business deeply, drives execution, and utilizes modern tools to simplify complexity and guide the company forward.