Cardinal 在从文档智能转型后推出 AI 营收平台

Why it matters

Cardinal shows where AI startup competition is moving as software gets cheaper to build: founders are productizing distribution, with sales execution becoming the scarce advantage.

Jianna Liu (@jianna_liu) and Devishi "Devi" Jha (@DevishiJha) launched Cardinal as an AI revenue platform on Thursday, expanding an outbound sales product that the repeat Y Combinator founders built after selling their previous startup to Johnson Controls.

https://x.com/jianna_liu/status/2082858765393797625?s=46

Liu said in a thread on X that companies including Deepgram, Giga and Mintlify use Cardinal to "design how they sell." Cardinal finds prospects, monitors buying signals and social activity, generates personalized messages, coordinates outreach and retargets engaged leads with ads. The software can run campaigns across email, LinkedIn and X, with teams managing and reviewing the work in Slack. (x.com)

此次发布将 Cardinal 早期以精准外呼为主的定位扩展为对整个营收栈的更广泛主张。五个月前，Cardinal 在 Y Combinator 的发布目录 上将自己描述为一个 AI 平台，替代了 10 多种销售工具以及一名专职 go-to-market 工程师的工作。Cardinal 当时表示，该产品已为超过 40 家 YC 公司运行外呼。(ycombinator.com)

Cardinal 目前的产品覆盖潜在客户发现、账户过滤、共享上下文研究、个性化片段、多渠道序列以及来自网站访问、产品注册、社交互动和广告的入站信号捕捉。其网站还宣传了 CRM 集成、团队分析和来自 Cardinal 的实操支持。(ycombinator.com)

From energy software to sales agents

Cardinal is the second startup Liu and Jha have built together. Jha studied computer science at Harvard, while Liu graduated from MIT with a computer science degree and previously worked on software teams at Meta, DoorDash and Nvidia. Their first company, Leafpress, used AI to collect and analyze utility data for corporate energy and carbon reporting. The founders took Leafpress through YC's Summer 2023 batch and later sold it to Johnson Controls, according to their current YC profiles. (ycombinator.com)

出售 Leafpress 塑造了创始人下一款产品的方向。Cardinal 表示，Liu 和 Jha 通过冷外呼赢得了 Lindt 和 Honeywell 等客户，这让他们亲身体验并组装了现在为其他公司打包的潜客挖掘、数据丰富、信息传递和序列化系统。(ycombinator.com)

通往周四发布的路径包含一次显著的产品转向。Cardinal 最初作为文档智能 API 公开发布，这是在 Liu 和 Jha 在 Leafpress 从公用事业账单中提取结构化信息时遇到困难之后构建的。创始人表示，该 API 使用两个微调模型将复杂文档转换为 Markdown 和代码，同时保留表格、注释和布局。(ycombinator.com)

该文档产品仍可通过 Cardinal 的 API 文档 访问，而 Cardinal 的主网站已将销售宣传转向外呼自动化。Revenue Platform（营收平台）的标签为创始人提供了超越潜客挖掘和序列化，进入广告、入站资格评估、内容监控和销售运营的空间。(docs.trycardinal.ai)

Cardinal 还比通常的自助式软件供应商拥有更重的服务层。在其公司页面上，Cardinal 将其模型描述为 "half-software, and half-service." Liu、Jha 和他们的团队帮助客户连接现有系统，围绕经过验证的销售策略配置代理，并通过 Slack 监控绩效。此交付模型可以加速没有 go-to-market 工程师的小团队的采用，但也使增长依赖于 Cardinal 招募能够设计和管理销售系统的人员的能力。(trycardinal.ai)

Cardinal's traction claims

Earlier this year, Liu said Cardinal had raised more than $5 million from Y Combinator, Audacious and operators including Ironclad CEO Geoff Donaker. Cardinal has not broken out the financing into named rounds. (linkedin.com)

Liu also said more than 50 YC companies were using Cardinal and that its campaigns had generated millions of dollars in customer pipeline. Jha separately said Cardinal reached a seven-figure annual recurring revenue run rate in roughly three months with its two founders. Those figures are company-reported, and pipeline measures potential deals rather than booked revenue. (linkedin.com)

创始团队正在招聘工程师、一名设计师以及一名在旧金山的创始级客户执行。Cardinal 的职位列表为一些早期岗位提供高达 1% 的股权，这反映出产品有多大程度上仍依赖将 Liu 和 Jha 目前亲自执行的销售工作进行编码化。(ycombinator.com)

Liu 的发布论点很直接：AI 已经减少了构建软件所需的时间，剩下的限制是分发。Cardinal 是创始人尝试将他们在 Leafpress 使用的外呼方法变成可重复用于其他初创公司的代理。其更广泛的营收平台定位提高了赌注，使 Cardinal 面临既有 CRM 和销售参与系统的竞争，也面对已经占据相同工作流部分的新工具，如 Clay、Apollo 和 Instantly。