从上下文归集到辅助决策From context to decision support
YouNavi 的价值不只是把一段对话压缩成会议纪要，而是把分散的会议、沟通、录音和资料归集为可持续理解的上下文，再从中识别分歧、风险、变化与下一步行动。我负责覆盖信息进入系统、被可靠处理，到用户获得可追溯洞察的关键链路。
YouNavi does more than compress a conversation into meeting notes. It brings fragmented meetings, conversations, recordings, and materials into context that can be understood over time—then surfaces divergence, risk, change, and next actions. My work spans the critical path from bringing information in, to processing it reliably, to delivering traceable insight.
- 上下文归集Context collection处理录制、转写、分段上传、状态恢复和结果衔接，让会议、沟通和录音等分散信息连续进入同一分析流程。Handled recording, transcription, segmented uploads, state recovery, and result continuity so fragmented meetings, conversations, and recordings can enter one analysis flow.
- 洞察与辅助决策Insight and decision support把复杂的后台状态和 Agent 分析转化为明确反馈，帮助用户从表层总结进一步识别分歧、风险、变化趋势与可执行的下一步。Translated complex system states and Agent analysis into clear feedback, helping users move beyond surface summaries to identify divergence, risk, changing signals, and actionable next steps.
- 多渠道同步Multi-source sync支持飞书、钉钉、腾讯会议、讯飞、Plaud、TicNote、GetSeed 与苹果设备等至少八类公开来源，让分散的对话数据进入统一上下文。Supports at least eight public source categories—including Feishu, DingTalk, Tencent Meeting, iFlyrec, Plaud, TicNote, GetSeed, and Apple devices—so fragmented conversation data can enter one context.
- 桌面端可靠性Desktop reliability覆盖 macOS / Windows 发布、更新、安装、健康监控和日志定位，让功能不仅能开发出来，也能稳定到达真实用户。Covered macOS and Windows releases, updates, installation, health monitoring, and diagnostics so features reliably reach real users.
从问题到上线From problem to launch
我最有成就感的是参与搭建 YouNavi 的 AI 辅助决策方案。它解决的不是“记录一次会议”，而是把会议、沟通、录音和资料汇集成长期可用的上下文，帮助用户发现分歧、风险和下一步行动。
The work I am proudest of is helping build YouNavi’s AI decision-support workflow. It goes beyond recording a meeting: it brings meetings, conversations, recordings, and materials into durable context so users can identify divergence, risk, and next actions.
作为第一位核心研发，我从真实使用和故障反馈定位问题，完成多渠道接入、洞察分析、桌面端发布和持续迭代。产品已支持至少八类公开对话/录音来源；我以“从采集到形成可行动建议”的完成情况、异常恢复和用户反馈，持续评估方案效果。
As the first core engineer, I worked from real usage and failure feedback through multi-source integration, insight workflows, desktop releases, and continuous iteration. The product supports at least eight public conversation and recording source categories; I assess its ongoing impact through the completion of the path from collection to actionable guidance, recovery from exceptions, and user feedback.
我对这类产品的理解How I think about this product
会议只是入口，不是终点。产品真正需要解决的是：如何把不断累积的对话上下文转化为可追溯、可校准、可继续行动的判断依据，而不仅是生成一份总结。工程上的可靠性、隐私边界与产品上的信任感，是同一件事的不同侧面。
Meetings are an entry point, not the destination. The real challenge is turning accumulated conversation context into a basis for judgment that is traceable, calibratable, and actionable—not merely generating a summary. Engineering reliability, privacy boundaries, and product trust are different sides of the same problem.