Project team

Jade Zhao, 2 Engineers


Product strategy, Research, UX/UI design,Design system


6 Weeks (2023)

DocHero is a GPT-Powered academic writing assistant. With over 75,000 writing sessions empowered to date, this case study unfolds our evolution from a simple translation tool to a robust platform acclaimed for its high retention rates


Amidst the global buzz of ChatGPT, China's huge user base was sidelined due to firewall restriction. DocHero was founded to fill this gap

Staus Quo

Initial success

Starting as a rephrasing tool, DocHero quickly gained seed users

2K Weekly active users

6K Registered users


Following the initial surge of success, we've hit a growth bottleneck. In a language rephrasing market teeming with fierce competition, the pressing question emerges:
HMW strategically navigate this landscape to carve out DocHero's unique edge?

who are users
Step 01 - categorize our users

More than 80% of users seek help for academic writing

By analyzing the types of text processed on our website, we've been able to categorize our users and understand their backgrounds. Notably, we discovered that more than 80% of our users visit our site for assistance with academic-related writing

81% Academic writing

11% International bussiness

9% Others

International students

I'm looking  for translations that are academically sound and closely resemble human writing, which could pass AI-detector

PHDs and researchers

I require linguistic precision and rigor for my academic papers, ensuring they meet the publication standards

Step 02 - Define persona

Zooming in to understanding the academic writers

After conducting a survey among our registered users, we've identified two key personas: international students, and PhDs and researchers. Both encounter language rephrasing challenges, yet their needs subtly differ

Define opportunities

Deep Dive: Identify the opportunities

To comprehensively analyze the user's journey in academic writing and precisely identify areas of challenge, I conducted 12 1-hour research sessions


Tool fragmentation disrupts academic workflow

Academic writing involves numerous tools that don't seamlessly integrate, leading users to hop between platforms for reading, note-taking, and writing


Language and context barrier in reading and writing

A strong need for tools that aid in comprehending content, enhancing vocabulary, refining sentence structures, and grasping subtle academic nuances


Formatting woes & Citation friction

Strict formatting guidelines for different conference and journalsGenerating accurate citations from their readings adds another layer of complexity and friction

Proof of concept testing


Based on opportunities identified in research, our team efficiently brainstormed and narrowed down to two key concepts

Prototype used for POC interview

Concept 01

Analyze research papers with AI

Concept 1 automates key research tasks, such as summarizing papers, extracting data, and synthesizing findings. These tasks are performed on our platform, with the results accessible through a plug-in in existing writing applications, ensuring seamless integration with users' current workflows


Users prefer solutions that complement, rather than replace, their existing habits. By enhancing the efficiency of the reading and synthesis phases with minimal disruption to user behavior, we expect our product to add significant value

Prototype used for POC interview

Concept 02

Academic writing assistant

Concept 2  enhances academic writing by using GPT for rephrasing and offering other writing aids. It allows users to upload, read, and seamlessly cite papers on our platform, creating a unified environment for reading and writing


There's a growing need for a comprehensive, AI-powered writing platform. By offering an integrated, all-in-one solution, we could deliver a streamlined and innovative writing experience, enhancing efficiency and creativity for users

What we want to learn

In our proof of concept interviews, we focused on two main questions:


Reading vs. writing

Identifying which area poses a greater challenge for our users and requires more support – reading and synthesizing information (Concept 1) or the writing and organizing process (Concept 2)


Plug-in vs. integrated platform

Evaluating user tendencies towards either integrating our tool as a seamless plug-in into their existing workflows, or preferring a shift to an all-encompassing, standalone platform


Concept 2 (academic writing assistant) is the winner. Here is why:

Divergent needs in the reading phase

Our two distinct personas with varying needs during the reading phase. International students found  features like summarizing and data extraction extremely beneficial. However, PhDs and professors did not value these features as highly, expressing a greater need for assistance in discovering related papers, which falls outside our team's technical expertise

Strong interest in AI-assisted writing

Both groups showed a significant interest in AI-assisted writing features.
They were particularly attracted to the idea of language rephrasing, formatting assistance, citation management, and plagiarism checking all in one place.

Mixed responses to adopting a new writing platform

Some users were open to the idea, especially if it meant a substantial improvement over their current tools.
Others preferred to complete the main body of their writing elsewhere, using our platform primarily for rephrasing and formatting tasks, indicating a gap in the market for an integrated solution that can complement existing platforms.

North star vision

North Star

Drawing from insights gained during the Proof of Concept interviews, I've formulated a north star vision to align our team around a cohesive product strategy


Design system

To enhance future development and ensure a unified product language, our team thoroughly evaluated multiple open-source design systems. We decided on Ant Design as the foundational framework for our visual system.
After extensive research and experimentation with different visual elements, we eventually landed on following approach that strongly aligns with our brand ethos, encompassing Intelligence, Trust and Reliability, and Vibrancy


To give our product more personality and better connect with younger users, we created our own IP (intellectual property)


Where we are today

Staus Quo


When we launched the test to all members we had over 75000+ writing session conducted in the first 8 weeks


Weekly active users grow from 2K+ to 6K+


Retention rate increased from 5.6% to 14.6%


Subscribers grow from 150+ to 200+