📡 Idea Radar 每天 8:00 / 18:00 推送 · 18:00 生成精选
reddit · SideProject · 52 分 ← 返回列表

I built an AI agent that gets confused by your SaaS on purpose

I kept noticing the same issue: founders build onboarding with full product context, while new users have none. So I built Botsona: AI personas that attempt real flows in a browser and show where they hesitate, fail, or take the wrong path. The question I’m trying to validate: is this genuinely useful before spending time on real user testing, or is it mostly expensive AI theater? Keep the video focused on one failed task and one resulting insight—not a brand montage. That

五维评分

付费意愿
45
需求真实度
40
个人可落地
75
竞争空间
45
新鲜度
70

AI 分析

痛点:SaaS 创始人用完整产品上下文设计 onboarding,而新用户没有上下文,导致真实使用中的犹豫、失败和错误路径难以在早期发现。

目标用户:SaaS 创始人、独立开发者、产品经理、UX 研究员和增长团队。

证据:The question I’m trying to validate: is this genuinely useful before spending time on real user testing, or is it mostly expensive AI theater? ... Can anyone try and give honest feedback?

变现:按测试运行次数或坐席订阅收费,例如免费 1 次体验,之后 $29-$99/月;也可按浏览器自动化分钟数或生成报告数量计费。

MVP:用 Playwright + LLM 视觉模型做一个浏览器代理,给 3 类预设人格(新手、急躁用户、怀疑者),跑通一条注册/激活流程,录制录屏并自动输出卡点时间轴、失败步骤和改进建议。

竞品:Maze、UserTesting、Lyssna、PlaybookUX、Hotjar、FullStory、Synthetic Users、Browser Use

摘要:作者发布 Botsona,用 AI 人格在浏览器中故意模拟新用户走 SaaS 流程,定位 onboarding 测试,展示用户在哪里犹豫、失败或走错路径。帖子主要寻求验证和反馈,没有付费意愿或收入数据。竞品包括 Maze、UserTesting 等,独立开发者可做轻量 MVP,但真实需求仍需验证。