来自YC F26批次的初创公司Pomerado近日在Hacker News发布Launch HN帖子,正式推出其核心产品——一个能够将任意网站在少于1小时内转化为确定性、自愈型MCP(模型上下文协议)API的平台 1。该公司由具备McKinsey和Scale AI背景的创始人Alan和Akshay创立,旨在解决AI智能体在与缺乏API的传统网站交互时面临的低效与不稳定问题 1。通过该平台,开发者可以摆脱对传统浏览器自动化的依赖,利用首次运行建立的模板生成可复用的确定性API,从而提升智能体执行任务的可靠性 1。
在技术实现上,Pomerado采用了一套独特的自动化工作流:首先由代理浏览网站并执行任务以记录数据,随后由Guardian模型审查这些数据并生成通用API,最后通过维护作业自动修复失败的步骤 1。相比传统的计算机使用(Computer Use)方案,经Pomerado生成的MCP调用通常在5至30秒内返回结果,具有成本更低、速度更快的优势 1。该产品不仅支持认证和双因素身份验证(2FA)处理,还已应用于个人助理预订航班、连接遗留经销商系统以及健身教练准备训练计划等场景 1。目前,Pomerado提供了MIT许可证的开源代码以及免费托管版本,供开发者和企业用户接入使用 1。
Pomerado, a startup from the Y Combinator Fall 2026 batch, has launched a new platform designed to transform any website into a deterministic and self-healing Model Context Protocol (MCP) interface 1. Founded by Alan and Akshay, who have backgrounds at McKinsey and Scale AI respectively, the tool aims to solve the inefficiency and instability associated with having AI agents interact with traditional websites that lack APIs 1. The core value proposition lies in its ability to generate these MCP APIs in less than an hour, offering a faster and more cost-effective alternative to browser automation or computer use methods 1.
The platform’s workflow involves an agent first browsing the website to execute a task, after which the data is recorded and reviewed by a "Guardian" model to create a reusable, generic API 1. Subsequent maintenance jobs automatically fix any failed steps, ensuring reliability over time 1. Compared to standard computer use approaches, calls generated through this system typically return results within 5 to 30 seconds 1. Pomerado supports complex features such as authentication and two-factor authentication (2FA) processing, making it suitable for various applications including personal assistants booking flights, connecting legacy dealer systems, and helping fitness coaches prepare training plans 1. The company offers both an MIT-licensed open-source version available on GitHub and a free hosted option via their web application 1.
评论
还没有评论,欢迎留下第一条。