展示HN:Sentient – 基于人工智能的客户反馈分析,准确率达到95%

2作者: iedayan0316 天前原帖
我们开发了Sentient,旨在自动提取客户反馈中的主题和情感,无需手动配置。该系统能够在亚秒级的时间内处理文档,同时保持95%的准确率。 Sentient基于Next.js 14和FastAPI构建,使用经过微调的OpenAI模型,这些模型是在数百万条反馈示例上训练而成。其主要功能包括自动客户细分、多格式支持(PDF、DOCX、CSV)以及实时处理。 核心见解是:大多数工具分析的是单个评论,而不是理解客户旅程和行为模式。Sentient通过多阶段的AI分析将反馈情感与商业结果相连接。 技术方法采用主题检测、情感识别和智能缓存,以实现企业级性能。 您可以在以下链接尝试:<a href="https:&#x2F;&#x2F;data-decoder.vercel.app&#x2F;" rel="nofollow">https:&#x2F;&#x2F;data-decoder.vercel.app&#x2F;</a>
查看原文
We built Sentient to automatically extract themes and sentiment from customer feedback without manual configuration. The system processes documents in sub-second time while maintaining 95% accuracy.<p>Built with Next.js 14 and FastAPI, using fine-tuned OpenAI models trained on millions of feedback examples. Key features include automatic customer segmentation, multi-format support (PDF, DOCX, CSV), and real-time processing.<p>The core insight: most tools analyze individual reviews instead of understanding customer journeys and behavioral patterns. Sentient connects feedback sentiment to business outcomes through multi-phase AI analysis.<p>Technical approach uses theme detection, emotion recognition, and intelligent caching for enterprise-grade performance.<p>Try it at: <a href="https:&#x2F;&#x2F;data-decoder.vercel.app&#x2F;" rel="nofollow">https:&#x2F;&#x2F;data-decoder.vercel.app&#x2F;</a>