reddit · SideProject · 10-26 21:50 51 分 ← 返回列表

I got tired of "Because you watched X" being useless, so I built a site that explains its recommendations

Hey everyone, Streaming recommendations have gotten completely stale for me. Whenever I finish something that really lands, like a tight psychological thriller, an intense character study, or a story with a very specific, gritty tone, the "Because you watched X" row is almost always useless. It just pushes the same generic genre tags or whatever licensed title the platform wants to promote that month. I was constantly spending 45 minutes doom-scrolling three different apps and digging thr

五维评分

付费意愿

需求真实度

个人可落地

竞争空间

新鲜度

AI 分析

痛点：流媒体平台的“Because you watched X”推荐高度同质化，用户想看调性、叙事手法精准匹配的作品时，只能跨多个 App 和多年前的 Reddit 帖子里翻找近 45 分钟。

目标用户：重度影视、剧集与动漫观众，尤其是对叙事结构、氛围调性有明确偏好的影迷

证据：I was constantly spending 45 minutes doom-scrolling three different apps and digging through years-old Reddit threads just trying to find something that scratched the exact same itch. / Free and zero friction: No paywall, no account needed to browse.

变现：短期靠 JustWatch、流媒体平台的观看渠道联盟分佣；中期加入付费档：跨平台收藏同步、调性标签自定义、片单导出和 API；长期可向发行方或小众流媒体出售推荐位。

MVP：用 LLM 为 800 部高口碑影视生成“主题、叙事手法、氛围调性”标签并做向量相似度图谱，接 TMDB 和 JustWatch 的元数据与观看渠道，前端只做两件事：输入片名出关系图，以及为每条连线写一句人话解释理由；单页 Web，两周可上线。

竞品：Taste.io、Letterboxd、JustWatch、Reelgood、Netflix 自家推荐算法、ChatGPT 找片

摘要：作者受够了流媒体“Because you watched X”的无用推荐，为了找同调性的片子要在三个 App 和旧 Reddit 帖间翻 45 分钟，于是做出 Tangent：以主题、叙事手法、创作基因把影视动漫连成“地图”而非列表，并用人话解释每条推荐的理由，还整合了 20 国的观看渠道信息。痛点真实且已被单人验证可交付，但作者明确做免费无账号、无付费墙，帖中没有任何付费意愿或收入数据，且影视发现与推荐赛道竞品非常密集。