reddit · SideProject · 10-11 20:50 54 分 ← 返回列表

I built an app that asks several AIs at once. How do you decide which answer to trust?

I kept getting different confident answers from different AI models, and I couldn’t tell which one to believe. Pasting the same prompt into several tabs got old fast, so I built CouncilAI. It sends one question to several models and gives you three ways to read the results: Council: the models answer independently, then a judge combines them into one answer. You can still read every individual response. Sequential: one model drafts, another critiques, a third refines

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：不同AI模型给出自信但矛盾的答案，用户难以判断该信任哪一个，逐个粘贴到多个标签页效率低。

目标用户：同时使用多个大模型的知识工作者、开发者、研究者、内容创作者等AI重度用户。

证据：I kept getting different confident answers from different AI models, and I couldn’t tell which one to believe. Pasting the same prompt into several tabs got old fast, so I built CouncilAI. It’s a one-time purchase on the Microsoft Store, Windows only for now.

变现：一次性买断/付费下载；后续可做Pro版、团队版、订阅制、API额度包或企业私有部署。

MVP：2周内做一个BYOK桌面端：输入一次问题并发给Claude/GPT/Grok/OpenRouter，提供Compare、Council、Sequential三种模式，本地保存历史，支持导出Markdown/JSON；先不做账号和支付，用买断或License key验证付费。

竞品：TypingMind、ChatHub、Msty、Poe、OpenRouter Chat、Perplexity、You.com

摘要：作者发布CouncilAI，把同一问题发给多个AI模型，并用裁决、顺序、对比三种模式辅助判断可信答案。产品采用自带API key、一次性买断、Windows Microsoft Store发布。该方向痛点多模型用户存在，但竞品密集，且帖子未展示销售或付费用户数据。