atypica.AI

An evolution journey from user discovery to step-by-step research, finally becoming a conversational research assistant 

import { fetchUserChatById } from "@/data";
import { authOptions } from "@/lib/auth";
import { getServerSession } from "next-auth/next";
import { forbidden, redirect } from "next/navigation";
import { Metadata } from "next/types";
import { StudyPageClient } from "./StudyPageClient";

export async function generateMetadata({
  searchParams,
}: {
  searchParams: Promise<{ id?: string }>;
}): Promise<Metadata> {
  const id = (await searchParams).id;
  if (!id) {
    return {};
  }
  const studyUserChatId = parseInt(id);
  const studyUserChat = await fetchUserChatById(studyUserChatId, "study");
  return studyUserChat.title ? { title: studyUserChat.title } : {};
}

export const dynamic = "force-dynamic";

export default async function StudyPage({
  searchParams,
}: {
  searchParams: Promise<{ id?: string }>;
}) {
  const { id } = await searchParams;
  if (!id) {
    redirect("/");
  }
  const studyUserChatId = parseInt(id);
  const studyUserChat = await fetchUserChatById(studyUserChatId, "study");

  const session = await getServerSession(authOptions);
  if (!session?.user) {
    redirect("/auth/signin?callbackUrl=/study");
  }
  if (studyUserChat.userId !== session.user.id) {
    forbidden();
  }

  return <StudyPageClient studyUserChat={studyUserChat} readOnly={false} replay={false} />;
}
atypica@localhost:~

git log --oneline | head -3

ee7febb update headers and usermenu, add pricing and study history on page header
044ebf3 add study list page
1ebfa76 fix lint issues

cat README.md | grep -i "atypica"

# atypica.AI

atypica.AI is a user research platform for subjective world modeling, driving research processes through large language models to help researchers quickly gain insights and generate research reports.

Welcome to submit Issues and PRs to improve atypica.AI together!

./storytell.sh --format timeline

Telling the evolution story of atypica.AI...

$ ./story --highlight "journey"

The story of atypica.AI began in early March 2025, when we faced a problem:

"User research takes too long. From finding target users, conducting interviews, to organizing reports, it usually takes weeks or even months. Could we use AI to accelerate this process?"

With this question, we began more than two months of rapid iterative development, evolving four different stages of product forms, each more powerful and user-friendly than the previous one.

Version One: User Discovery Tool 2025.3.8-3.13

It all started with a simple feature: using AI to search and analyze Xiaohongshu user data, automatically generating user personas.

Product Form
  • Single scout page where users input research objectives
  • AI automatically searches and analyzes relevant user data
  • Generates and saves user personas

"The first time I saw AI automatically collect user information from the internet and generate personas, I couldn't believe my eyes. This process usually takes me a whole day." — Early test user

scout.tsx
import { fetchUserChats } from "@/data";
import { Message, useChat } from "@ai-sdk/react";

export function ScoutChat() {
  const {
    messages,
    input,
    handleSubmit,
    status,
  } = useChat({
    api: "/api/chat/scout",
    body: {
      scoutUserChatId: scoutUserChatId,
    },
  });

  return (
    <div>
      <div>
        {messages.map((message) => (
          <ChatMessage
            key={message.id}
            role={message.role}
            content={message.content}
          />
        ))}
      </div>
      <form onSubmit={handleSubmit}>
        <textarea
          value={input}
          placeholder="Enter your research topic..."
        />
      </form>
    </div>
  );
}
Milestone: March 12, 2025 - First successful automatic user persona generation from Xiaohongshu data

Version Two: Step-by-Step Research Process 2025.3.14-3.19

Building on the success of the first version, we expanded the feature boundaries and built a complete user research process, with AI assistance and guidance at every step.

Product Form
scout
→
personas
→
analyst
→
interview
→
report
  • Multi-page step-by-step process with detailed guidance at each step
  • Researchers complete one task per page before moving to the next step
  • AI assists at every stage, reducing manual operations
  • Research results transfer seamlessly between steps

"The step-by-step design is very intuitive. I just need to follow the process step by step to complete the entire research. The guidance prompts on each page are also very thoughtful, making it easy for a novice like me to get started." — Market Researcher

Analyst

Research topic creation & management

Interview

AI-driven user interview simulation

Report

HTML report generation & rendering

Milestone: March 15, 2025 - Implemented complete step-by-step research process, reducing user operation burden through guidance

Version Three: Conversational Research Assistant 2025.3.20-3.27

Finally, we achieved a revolutionary breakthrough: integrating the entire research process into a single conversational interface, where users only need to chat with the Study Agent, and all steps are completed automatically in the background.

Product Form
  • Single conversational interface for complete research workflow
  • Split-screen design: left conversation, right tool console
  • Real-time display of AI execution process and results
  • No page jumping required, one-stop research completion

"This is simply magic! I just need to tell the AI what I want to research, then I can watch it automatically complete all the work, from finding users to generating reports, everything happens in real-time before my eyes." — Product Manager

study.tsx
Chat Interface
I want to understand the factors that influence young mothers' choice of baby formula
AI
I will help you research the key factors influencing young mothers' choice of baby formula. I will automatically perform the following steps:

1. Find target user groups
2. Generate user personas
3. Design and conduct interviews
4. Analyze results and generate reports
Tool Console
$
scoutTaskChat
Searching Xiaohongshu for discussions by young mothers about baby formula...
$
savePersona
Saved 3 typical user personas:
- 25-year-old first-time mother, focuses on safety
- 28-year-old mother of two, values cost-effectiveness
- 32-year-old working mother, prioritizes convenience
$
interview ●
Conducting interviews...1/3 completed
Milestone: March 20, 2025 - Launched unified conversational interface, implementing study agent coordination of other agents work model

Version Four: Commercialization & Platform Optimization 2025.3.28-5.12

After perfecting the core research functions, we turned to commercial operations and global deployment, adding features such as subscriptions, token billing, multi-language support, and more.

Product Form
  • Complete SaaS service with free and paid tiers
  • Multi-language interface and global deployment support
  • Token consumption statistics and recharge process
  • Research history and learning lists
  • Advanced interview project functionality

"The token billing model is very fair, I only pay for the research I actually use. Moreover, I can check consumption statistics at any time to understand the cost of each research project." — Enterprise User

Tokens

Token consumption stats & billing

i18n

Multi-language support & global deployment

Subscription

Pro subscription & permission management

Milestone: April 20, 2025 - Launched complete subscription and token billing model, achieving commercial operation

$ ./analyze --type "evolutionary_leaps"

01 操作方式的演进

  • v1 多步骤手动操作
  • v2 分步引导式操作，减少用户负担
  • v3 对话式一站操作，AI自动执行全流程
  • v4 商业化一站式研究平台，全球化部署

02 界面形态的变化

  • v1 单功能页面
  • v2 多页面串联流程
  • v3 单页面分屏设计，左对话右执行
  • v4 统一全局导航，多语言支持，移动响应式设计

03 AI协作模式的升级

  • v1 单一AI助手
  • v2 多个独立AI功能点
  • v3 指挥官模式，Study Agent统筹协调
  • v4 多模型策略，根据任务智能切换不同LLM

$ ./analyze --type "commercial_model"

04 商业模式的演进

  • v1-v3 功能验证阶段，无商业模式
  • v4 引入令牌经济与订阅制，平衡用户体验与可持续发展

主要收入来源

按需付费令牌
充值即用
专业版订阅
月付/年付
企业解决方案
定制化

$ ./architecture --visualize

多Agent协作系统

Study Agent

全流程协调者，引导用户明确研究需求

Scout

发现目标用户

Interviewer

执行专业访谈

Persona

模拟用户回答

工具集成
reasoningThinking

深度思考分析工具

interview

自动化访谈管理

scoutTaskChat

用户发现与画像构建

generateReport

报告生成与渲染

$ ./conclusion

在短短两个多月内，atypica.AI完成了从单一功能工具到商业化研究平台的蜕变，核心演进是从"多步骤操作"到"分步引导式操作"，再到"对话式一站操作"，最终成为"商业化一站式研究平台"。

最新版本不仅保留了简洁的用户体验，还实现了商业可持续性，支持全球化部署和多语言环境，通过令牌经济和订阅模式为各类用户提供了强大而直观的主观世界建模工具。

"atypica.AI的发展历程体现了我们团队的核心理念：通过技术创新持续降低用户研究的门槛，让每个人都能轻松获取深度洞察，同时建立可持续的商业模式，确保产品能持续为用户创造价值。" — 项目负责人

atypica@localhost:~

cat roadmap.txt

未来计划:
- 扩展AI模型选择，优化不同任务的性能表现
- 增强跨语言和跨文化研究能力
- 研发更高级的数据可视化工具
- 添加企业级团队协作与权限管理
- 构建开发者API，实现自定义研究工具
- 优化令牌使用效率，降低研究成本

git commit -m "Continue the journey..."

[main 3a7b56c] Continue the journey...
Ready for next chapter