您是如何处理客户的停机补偿的？（云服务 / 大语言模型基础设施）
大家好！
我们正在研究“即时停机保护”，需要验证这是否是一个真实的问题，还是仅仅是我个人的感受。<p><i>有两个问题</i>：
1. 如果你们在销售B2B SaaS / API产品：当你的基础设施出现故障（或者你的上游服务提供商——AWS、GCP、Supabase、Stripe、Cloudflare等）时，你们是如何处理客户赔偿的？<p>客户真的会申请SLA信用吗？这是一个手动支持工单的过程吗？有没有人实现了自动化？<p>2. 如果你们在使用LLM API（如OpenAI、Anthropic等）：当模型出现故障或受到速率限制，导致你的产品无法正常工作时，客户会期待赔偿吗？你们目前是如何处理的？<p>背景：我们刚刚推出了针对航班延误的即时自动赔付（<8分钟到客户的UPI，无需索赔表格，与印度最大的航空公司合作）。我们正在探索同样的基础设施是否适用于云/SaaS停机。<p>我们的论点是：一个即时自动到账的$100信用，对客户的价值远高于一个需要客户主动申请并等待30天的$100信用。<p>我们会与所有填写此调查的人分享我们的发现。<p>如果你有2分钟的时间，非常希望你能在这里提供你的意见：https://tally.so/r/OD7Llg
查看原文
Hey all!
We are working on “Instant Downtime Protection” and need to validate whether this is a real problem or just me.<p><i>Two questions</i>:
1. If you're selling B2B SaaS / API products: When your infra goes down (or your upstream providers - AWS, GCP, Supabase, Stripe, Cloudflare, whoever), how do you handle customer compensation?<p>Do customers actually claim SLA credits? Is it a manual support ticket process? Does anyone automate this?<p>2. If you're building on LLM APIs (OpenAI, Anthropic, etc.): When the model is down or rate-limited and it breaks your product, do your customers expect compensation? How do you handle it today?<p>Context: We have just rolled out instant automatic payouts for flight delays (<8 min to customer's UPI, zero claim form with India’s largest airline). Exploring whether the same infrastructure works for cloud/SaaS downtime.<p>The thesis: a $100 credit that arrives instantly and automatically is worth way more to customers than a $100 credit they have to ask for and wait 30 days for.<p>We'll share findings with everyone who filled this out..<p>If you've got 2 minutes, would love your input here: https://tally.so/r/OD7Llg