供资
Worp Sciences Emergings From Secretth with 215M for AI-Powered Chemistry Labs 互联网档案馆的存檔，存档日期2013-12-20.
在 Google 上的首选来源中添加 Unite.AI
设计有前途的分子只是药物研究的开始。 还需要有人做，测试，然后把结果变成更好的下一个实验。 Warp Sciences打赌这个物理瓶颈是AI可以创造下一个主要优势的地方。
Warp Sciences已经从隐蔽中崛起，拥有2.15亿美元的资金，用于建立一个以美国为基地的化学合同研究组织，即CRO，结合了有经验的化学家，AI，和机器人。 公司由前OpenAI首席产品官凯文·韦尔(英语：Kevin Weil)领导，计划向生物技术和制药客户提供研究化合物，同时建设实验室，学习他们所进行的实验。
Lightspeed领导了融资，参与的有IVP,Mayfield,Lux,Premji Invest，以及Menlo Ventures等。 天使投资者包括布莱恩·阿姆斯特朗，迈克·克里格，迪伦·菲尔德，以及OpenAI的马克·陈和莎拉·弗里尔。
围绕AI的化学服务业务
Warp的起点是铅优化模拟合成：使分子变异，让研究人员在对药物候选者工作时能够比较它们。 这是一种可交付的具体服务，而不是AI模型将独立发明药品的承诺。
根据其客户服务页面，Warp专注于临床前小分子程序，以毫升至多克比例生产化合物。 该公司将其典型的工作描述为需要五至十五个合成步骤的细化分子。 这种范围使它处于研究团队需要完善的化合物制作和评价的反复周期之中。
华普还明确区分研究议程的所有权。 客户选择哪些分子来追求和保持对其药物程序的控制。 华普没有自己的专有药品管道。 其业务是提供这些团队使用的化学和实验基础设施。
这很重要 CRO必须提供可用的化合物，保持质量，并与客户科学家进行有效合作。 这些要求使Warp的自动化工作立即成为一项业务测试：它是否使服务更具生产力和可靠性。
从实验室结果到更好的模型
较长期的技术论文是实验和机器学习之间的反馈循环。 物理实验能产生关于什么是有效的证据。 AI系统从这些证据中学习，帮助指导随后的实验，并支持日益自动化的实验室工作。 更有能力的实验室然后可以产生更有用的证据。
在其创始人宣言Warp 称中，它计划捕捉成功的实验和失败。 除了客户工作外，它还打算开展专门的实验活动，为其模型生成数据。 随着更多工作流程自动化，有经验的化学家将领导实验室。
包括失败是方法的一个重要部分。 如果知道试验没有产生预期的结果，将有助于确定方法的限度，并告知下一步应进行何种试验。 为了改进模型，需要用足够的实验背景来记录这种信息，以区分非生产性的想法和执行中的问题。
机器人提供循环的物理侧面，而软件连接规划，执行，以及记录的结果。 在实验室旁边建造这些系统使Warp有机会围绕化学家实际进行的工作设计自动化。 建议的优势来自改进整个实验过程，而不是在保持实验室队列不变的同时加速分子设计。
Warp的广义野心是科学AI，可以帮助科学家对物理世界进行理性分析，设计信息实验。 其宣布商业起点仍然较窄：提供研究化合物并逐步自动化生产这些化合物所需的工作。
四个创始人化学、人工智能和业务
凯文·韦尔(Kevin Weil)，联合创始人兼CEO，最近担任OpenAI首席产品官和科学副总裁。 他的角色带来体验了将AI的进步转化为产品和周围的操作系统。
联合创始人和COO DJ Patil是第一位美国首席数据科学家，也是Deveted Health公司的创始执行官。 联合创始人兼首席科学官索尔·赖斯贝格以前曾在特尔莫托生物科学公司担任药用化学领导职务。 联合创始人和CTO Mano Paluri是Meta超级智能实验室的副总裁，也是FAIR的创始成员。
组合反映了前面的工作：构建AI系统，运行科学操作，以及交付化学客户可以使用。 Warp 称它的早期团队还包括来自创世纪的机器人体验，Skild AI和1X，以及Genenentech和Lawrence Livermore National Labs的科学专长。
为交付的化合物充电
Warp的定价方式是新增，其网站称提供每个大院的公司收费，价格与客户协商商定。 未生产化合物的，客户不支付。 这种安排与基于全时等同人员配置的CRO合同形成对比。
这一模式将收入与已完成的工作联系起来。 如果自动化降低了可靠制造复合物所需的努力，Warp可以改善自身的经济学，同时让客户在支出和产出之间有更直接的联系。 该公司将其服务定位为以境外定价为主的国内协作和控制，尽管它没有公布一般价格清单或比较成本结果。
它目前正在与少数设计伙伴一起工作。 这些关系涉及客户化学家和Warp的团队每周合作，使启动者能够根据积极的研究需要改进其服务和系统。
2亿1千5百万美元背后的处决挑战
这笔资金为WARP公司提供了大量支持，该公司必须建设实体能力和软件。 它的机会是将化学服务转变为学习系统：提供化合物，获取实验证据，改进模型，并利用这些改进来提高以后的工作效率。
该公告没有提供经过衡量的周转改进、基准测试型号或充分自主的实验室已经在运行的证据。 这些是组团和资本的独特里程碑。 实际的测试将是Warp能否反复提供质量，速度，以及它所前景的客户。
现在，公司最间接的选择是它开始的地方。 Warp从客户主导的化学计划开始，将AI的野心与不断出现的科学需求联系起来。 每个交付的化合物都提供了赚取收入和改善下一个化合物后面的实验室系统的机会。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
Funding
Warp Sciences Emerges From Stealth With $215M for AI-Powered Chemistry Labs
Add Unite.AI to your preferred sources on Google
Designing a promising molecule is only the beginning of drug research. Someone still has to make it, test it, and turn the results into a better next experiment. Warp Sciences is betting that this physical bottleneck is where AI can create its next major advantage.
Warp Sciences has emerged from stealth with $215 million in funding to build a U.S.-based chemistry contract research organization, or CRO, combining experienced chemists, AI, and robotics. Led by former OpenAI Chief Product Officer Kevin Weil, the company plans to provide research compounds to biotech and pharmaceutical customers while building laboratories that learn from the experiments they perform.
Lightspeed led the financing, with participation from IVP, Mayfield, Lux, Premji Invest, and Menlo Ventures, among others. Angel investors include Brian Armstrong, Mike Krieger, Dylan Field, and OpenAI’s Mark Chen and Sarah Friar.
A Chemistry Services Business Built Around AI
Warp’s starting point is lead optimization analog synthesis: making variations of molecules so researchers can compare them as they work toward a drug candidate. It is a concrete service with a deliverable, rather than a promise that an AI model will independently invent a medicine.
According to its customer services page, Warp is focused on preclinical small-molecule programs, producing compounds at milligram to multi-gram scale. The company describes its typical work as elaborated molecules requiring five to 15 synthetic steps. That scope places it inside the repeated cycle of making and evaluating compounds that research teams need to refine a program.
Warp also makes a clear distinction about ownership of the research agenda. Customers choose which molecules to pursue and retain control of their drug programs. Warp has no proprietary drug pipeline of its own. Its business is supplying the chemistry and experimental infrastructure those teams use.
That matters commercially. A CRO has to deliver usable compounds, maintain quality, and collaborate effectively with customer scientists. Those requirements give Warp’s automation effort an immediate operational test: whether it makes the service more productive and dependable.
From Laboratory Results to Better Models
The longer-term technology thesis is a feedback loop between experimentation and machine learning. Physical experiments produce evidence about what works. AI systems learn from that evidence, help guide subsequent experiments, and support increasingly automated laboratory work. More capable labs can then produce more useful evidence.
In its founders’ manifesto, Warp says it plans to capture both successful experiments and failures. Alongside customer work, it intends to run dedicated experimental campaigns to generate data for its models. Experienced chemists will lead the laboratories as more of the workflow is automated.
Including failures is an important part of the approach. Knowing that an experiment did not produce the desired result can help define the limits of a method and inform what should be tried next. For that information to improve a model, it needs to be recorded with enough experimental context to distinguish an unproductive idea from a problem in execution.
Robotics supplies the physical side of the loop, while software connects planning, execution, and recorded outcomes. Building those systems alongside the laboratory gives Warp the opportunity to design automation around the work chemists actually perform. The proposed advantage comes from improving the entire experimental process, rather than speeding up molecular design while leaving the lab queue unchanged.
Warp’s broader ambition is scientific AI that can help scientists reason about the physical world and design informative experiments. Its announced commercial starting point remains narrower: delivering research compounds and progressively automating the work required to produce them.
Four Founders Bridging Chemistry, AI, and Operations
Kevin Weil, co-founder and CEO, most recently served as Chief Product Officer and Vice President of Science at OpenAI. His role brings experience translating advances in AI into products and operating systems around them.
Co-founder and COO DJ Patil was the first U.S. Chief Data Scientist and a founding executive at Devoted Health. Co-founder and Chief Scientific Officer Sol Reisberg previously held medicinal chemistry leadership roles at Terremoto Biosciences. Co-founder and CTO Mano Paluri was a vice president at Meta Superintelligence Labs and a founding member of FAIR.
The combination reflects the work ahead: building AI systems, running scientific operations, and delivering chemistry customers can use. Warp says its early team also includes robotics experience from Genesis, Skild AI, and 1X, alongside scientific expertise from Genentech and Lawrence Livermore National Labs.
Charging for Compounds Delivered
Warp’s pricing approach adds another dimension to the funding story. Its website says the company charges per compound delivered, with pricing agreed in consultation with the customer. If a compound is not produced, the customer does not pay for it. Warp contrasts that arrangement with CRO contracts based on full-time-equivalent staffing.
That model connects revenue to completed work. If automation reduces the effort required to make a compound reliably, Warp could improve its own economics while giving customers a more direct connection between spending and output. The company is positioning its service around domestic collaboration and control at offshore pricing, although it has not published a general price list or comparative cost results.
Warp says it is currently working with a small number of design partners. Those relationships involve weekly collaboration between customer chemists and Warp’s team, giving the startup a way to refine its service and systems against active research needs.
The Execution Challenge Behind the $215 Million
The funding gives Warp substantial backing for a business that must build physical capacity as well as software. Its opportunity is to turn chemistry services into a learning system: deliver compounds, capture experimental evidence, improve the models, and use those improvements to make subsequent work more efficient.
The announcement does not provide measured turnaround improvements, model benchmarks, or evidence that a fully autonomous laboratory is already operating. Those are distinct milestones from assembling the team and capital. The practical test will be whether Warp can repeatedly deliver the quality, speed, and economics it is promising customers.
For now, the company’s most consequential choice is where it begins. By starting with customer-led chemistry programs, Warp is tying its AI ambitions to a recurring scientific need. Each delivered compound offers a chance to earn revenue and improve the laboratory system behind the next one.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.