Making Cities Awesome: Peregrine’s Nick Noone & Ben Rudolph

以下是原文的中文翻译： Peregrine 是一家人工智能公司，致力于利用技术提升城市、县和州的公共安全和社区福祉，同时坚决拒绝监控国家的概念。其核心使命是通过确保客观安全和主观安全感来让城市变得“更美好”，相信这种稳定性将开启无限可能性。 创始人 Nick 和 Ben 为这项使命带来了宝贵的经验。Nick 的背景包括运营 Palantir 的 SOCOM 部门，在那里他学会了“前沿部署工程”（FDE）——即深入了解客户问题，迅速取得成果，并优先考虑客户的成功。他指出，硅谷常误解 FDE，缺乏对复杂制度背景的同理心，并表现出傲慢。Ben 在联合国难民署和 Demagi 从事人道主义援助工作，在非洲和印度构建“最后一英里”医疗解决方案，这凸显了数据问题如何构成了社会挑战的基础，从而激发了他对技术影响力的关注。 他们的旅程始于深入剖析如何与美国城市合作的目标。在经历了多次拒绝后，他们于2018年2月通过主动向公共安全领域受人尊敬的人物 Brian Bubar 指挥官学习，获得了接触圣巴勃罗警察局的机会。这种对倾听和理解的重视对于建立信任至关重要。 Peregrine 在“削减警察经费”运动和社会动荡中启动，在一个两极分化的环境中航行。Nick 形容自己持守着“两种真相”——朋友们抗议警察，而他的团队却与执法部门合作。这一时期巩固了他们质疑假设、为常被“误解”的机构提供非政治性解决方案的承诺，即使这意味着工程师们不愿加入他们的努力。 Peregrine 的商业模式与 Flock 或 Axon 等数据收集公司形成鲜明对比。Peregrine 并非收集更多数据，而是专注于整合和利用客户*已经拥有*的零散信息。他们在现有系统之上提供一个安全、受管制的解决方案，以提高精确性和准确性。Peregrine 强调数据治理、权限控制和主权，坚称数据属于客户和社区，而非公司，旨在去中心化控制并防止形成“中央全景监狱”。 他们的应用程序展示了这种方法的力量。在佛罗里达州，人工智能帮助一个县将特定的三天天气模式与离岸流联系起来，为水上救援工作提供信息。对于犹太教堂威胁，人工智能的语义搜索识别出了关键词搜索遗漏的反犹模式。他们的“悬案代理人”处理数百GB的各种数据（视频、音频、PDF），以获取洞察，例如从零散的手机记录中识别嫌疑人的位置，最终促成了一起威斯康星州案件的无罪释放。 在内部，人工智能显著加速了软件开发，代理（agent）编写了大约90%用于数据集成的 Python notebooks。FDEs 充当“创新实验室”，创造独特的、客户专属的工具，如飓风模拟器或消防部门影响评估器，通常速度很快。这种“独立研发模式”赋予个体贡献者权力，在组织边缘培养创新，并为产品开发提供重要反馈。 在解决像面部识别这样的道德复杂性时，Peregrine 坚称，强制推行技术决策并非他们的职责。相反，他们向客户告知背景、法律考量和社区偏好，允许每个城市做出自己的知情选择——目前大多数选择不使用面部识别。 他们深入的客户参与传统上被认为是难以规模化的，但通过构建一个垂直整合、一流的技术栈，他们实现了规模化。这使他们能够以州、县和市级实体负担得起的价格点（每年低于一百万美元）提供定制化、复杂的解决方案，触及到以前服务不足的群体。 Peregrine 设想未来成为10,000个城市的机构记忆层，赋予他们管理数据同时保留其独特身份的能力。这一长期愿景建立在坚定不移的诚信、强大的道德指南和对管理责任的承诺之上，确保技术改善生活和社区，同时不施加不当权力。

Peregrine is an AI company dedicated to leveraging technology to enhance public safety and community well-being in cities, counties, and states, emphatically rejecting the concept of a surveillance state. Its core mission is to make cities "awesome" by ensuring both objective safety and a subjective feeling of security, believing this stability unlocks immense possibilities. The founders, Nick and Ben, bring formative experiences to this mission. Nick's background includes running Palantir's SOCOM unit, where he learned "forward deployed engineering" (FDE) – deeply owning a customer's problem, rapidly achieving outcomes, and prioritizing the customer's success. He notes Silicon Valley often misinterprets FDE, lacking empathy for complex institutional contexts and exhibiting ego. Ben's work in humanitarian aid with the UN Refugee Agency and Demagi, building last-mile healthcare solutions in Africa and India, highlighted how data problems underpin societal challenges, inspiring his focus on technology's impact. Their journey began with the goal of deconstructing how to work with American cities. After numerous rejections, they gained access to the San Pablo Police Department in February 2018 by offering to learn from Commander Brian Bubar, a respected figure in public safety. This emphasis on listening and understanding was crucial for building trust. Launching amidst the "defund the police" movement and social unrest, Peregrine navigated a polarized environment. Nick describes holding "two truths" – friends protesting police while his team worked with law enforcement. This period solidified their commitment to questioning assumptions and delivering apolitical solutions for often "misunderstood" institutions, even when it meant engineers were reluctant to join their efforts. Peregrine's business model starkly contrasts with data collection companies like Flock or Axon. Instead of gathering more data, Peregrine focuses on integrating and utilizing disparate information that customers *already own*. They provide a secure, governed solution atop existing systems to improve precision and accuracy. Emphasizing data governance, permission controls, and sovereignty, Peregrine maintains that data belongs to the customer and community, not the company, aiming to decentralize control and prevent a "central panopticon." Their applications demonstrate this approach's power. In Florida, AI helped a county link specific three-day weather patterns to rip currents, informing water rescue efforts. For synagogue threats, AI's semantic search identified antisemitic patterns missed by keyword searches. Their "cold case agent" processes hundreds of gigabytes of diverse data (video, audio, PDFs) to glean insights, such as identifying a suspect's location from scattered cell records, leading to an exoneration in a Wisconsin case. Internally, AI significantly accelerates software development, with agents writing about 90% of Python notebooks for data integration. FDEs act as "innovation labs," creating unique, customer-specific tools like hurricane simulators or fire department impact estimators, often rapidly. This "dark cave" approach empowers individual contributors, fostering innovation at the organizational fringes and providing vital feedback for product development. Addressing moral complexities like facial recognition, Peregrine asserts it's not their role to impose technological decisions. Instead, they inform customers about context, legal considerations, and community preferences, allowing each city to make its own informed choice – most currently opting against facial recognition. Scaling their deep customer engagement, traditionally seen as unscalable, was achieved by building a vertically integrated, first-class technology stack. This allows them to offer tailored, sophisticated solutions at a price point (under a million dollars annually) affordable to state, county, and city entities, reaching those previously underserved. Peregrine envisions becoming an institutional memory layer for 10,000 cities in the future, empowering them to manage their data while preserving their unique identities. This long-term vision is underpinned by unwavering integrity, a strong moral compass, and a commitment to stewardship, ensuring technology improves lives and communities without exerting undue power.