Stock Agent
Overview
As a product designer of the conversational AI assistant app Adot
I designed the Stock Agent feature to provide users with a personalized conversational experience related to stock investments. The Stock Agent was one of Adot's five key agents, which were the major features of Adot 3.0 version release. Since its initial launch in September 2024, the feature has been continuously improved and enhanced through ongoing updates.
I designed the Stock Agent feature to provide users with a personalized conversational experience related to stock investments. The Stock Agent was one of Adot's five key agents, which were the major features of Adot 3.0 version release. Since its initial launch in September 2024, the feature has been continuously improved and enhanced through ongoing updates.
I was responsible for designing in-app screens and dialogue guides, and collaborated with a senior product designer on conversation design.
Adot is a conversational AI service developed by SK Telecom
It serves as a smart virtual assistant capable of understanding natural language queries, providing accurate information, and supporting daily tasks such as scheduling, reminders, and recommendations.
Adot is designed to continuously learn and improve through user interactions, offering seamless integration with other SK Telecom services and third-party platforms.
Product Design
Project Management
Collaborated with a senior product designer, back-end and front-end developers, and UI designers
2024. 02 - Present
Figma, Confluence, Jira, Slack
Goal
To provide users with a personal AI stock agent capable of delivering professional advice and information related to stock investments
Background
To provide a professional AI agent based on reliable information sources, we selected stock investment as the focus area for the following two reasons:
The age distribution of Adot users closely aligns with that of stock investors.
Among Adot users, 68.6% have used other stock investment services or trading platforms within the past 30 days.
Process
1 Target Audience Selection
We selected moderate-involvement investors as the target audience for the following two reasons:
Since the onset of COVID-19, the number of individual investors has surged significantly. As a result, the majority of domestic individual investors are currently presumed to be either low-involvement or moderate-involvement investors.
They show strong interest and independence in stock investments, actively seeking information and opportunities in the market.
- Possesses a basic understanding of securities-related content, enabling comprehension of key concepts.
-
Demonstrates moderate interest in stock investments and seeks additional information to make more informed decisions.
- Requires investment-related insights and benefits from the information provided by the Stock Agent to meet their needs.
2 Analysis of Pain Points and Other Services
Pain Points
- Many users struggle with identifying the information they need and where to find it.
- They often learn about critical updates too late or remain unaware of them altogether.
Users face difficulties in:
-
comprehending complex data
-
identifying the causes of price fluctuations
- interpreting disclosure statements
Other Services
Service Name
Service Usage Behavior
Pain Points
NAVER
- Browsing investment options on subscribed blogs
- Checking stock prices
If they notice a significant price drop, they search for related news to understand the cause. - Reviewing disclosures
When the meaning of the disclosure is unclear, they turn to discussion boards for further exploration and clarification.
- Lack of push notifications for information other than prices and disclosures
- Need for additional searches to understand the meaning of provided information
YouTube
- Watching content from subscribed economic and stock-related YouTubers
- Searching for videos about market issues
- Covers only large-cap stocks and trending issues
- Takes a long time to digest the content
Toss
- Checking stock prices
If they notice a significant price drop, they search for related news to understand the cause. - Reviewing disclosures
When the meaning of the disclosure is unclear, they turn to discussion boards for further exploration and clarification.
- Lack of push notifications for information other than prices and news articles
* Toss is a leading financial app in South Korea, offering a wide range of services including investment management.
Other MTS* services
- Checking stock prices
If they notice a significant price drop, they search for related news to understand the cause. - Buying/Selling Stocks
- Must find all information on one’s own
- Information format difficult to consume on mobile (Bad UX/UI)
* MTS(Mobile Trading System) refers to a mobile application or platform that allows users to trade stocks and manage their investments directly from their smartphones.
3 Defining Value Propositions
To address the pain points mentioned above, we defined the value propositions of the Stock Agent as follows:
01
Users want to quickly access important updates about their favorite stocks
→ Provide the latest information on favorite stocks through proactive notifications
02
Users are not accustomed to complex information searches and prefer easily accessible data
→ Provide simplified and translated information, with the option to search effortlessly through conversations when needed
03
Support users in making confident investment decisions by ensuring access to consistent and reliable data
Prevent hallucination issues and enhance data accuracy by leveraging APIs and databases from securities firms
4 Conversational Feature Definition
We designed conversational features corresponding to each value proposition.
At the same time, we sought to create an agent that proactively provides users with potentially useful information, rather than just passively responding to their requests.
We also aimed to develop well-structured reactive scenarios to ensure users could search freely based on their specific criteria.
2) utilized the GPT-4o model and applied prompt engineering to create natural and concise messages.
3) incorporated visual components and buttons to make the information easily comprehensible at a glance.
Below are the features designed based on the objectives above, categorized by value proposition.
Timeliness
Providing the latest updates on favorite stocks and economic information through proactive notifications
Daily morning updates with a summary of U.S. market news and prices of the user's favorite stocks
"Good morning! Here’s a summary of the U.S. stock market:
Major tech companies experienced poor earnings, leading to a sharp drop in stock prices. Tesla fell by 12.33%, ...”
Summary of important disclosures and translations (for U.S. stocks) provided immediately upon release
"SK Telecom announced its dividend disclosure on July 10. If you hold SK Telecom stocks until July 19, 2024, you will receive the dividend payout on August 7. ..."
Providing information on IPO subscription schedules to assist users in making investment decisions
"A new IPO subscription schedule is available. How about adding it to your calendar so you don’t miss it?"
Convenience & Reliability
Offering a search function that provides accurate and easily accessible answers
Conversational Stock Attribute Search
"What was SK Telecom's revenue in 2023?"
"Show me the top 5 stocks with the highest operating profit."
"Which are the top 5 stocks with the highest dividends?"
Stock News Search
"Show me news related to Samsung Electronics."
Recommended Stocks & Themes
"Here’s a must-check theme I found in today’s Hana Securities report!"
5 In-app Screens Design
Providing all the information users want solely through conversations can be overwhelming, so I designed supporting screens to complement the experience. Focusing on helping users find the information they need in a concise and intuitive way, I designed the following pages.
A detailed page for each stock, where users can view price charts by period, related news, disclosures, relevant themes, associated YouTube content, and financial statements.
A page where users can easily view and edit their list of registered favorite stocks at a glance.
6 Dialogue Suggestions Design
To facilitate and encourage more engaging conversations between users and the Stock Agent and fully utilize its features,
we designed a system to provide dialogue suggestions.
In order to create dialogue suggestions that align with the conversation context, we extracted entities like stock names and attributes from the previous dialogue and used them to generate relevant suggestions for the user.
"What is HBM?"
"Compare Samsung Electronics and SK Hynix."
"Show me Samsung Electronics' operating profit for the last 3 years."
Outcomes & Refinements
The feature was successfully launched in September 2024, with 33,963 users utilizing the Securities Agent within the first month.
An analysis of user utterances revealed that loyal users often requested more complex and in-depth information.
To address this, we expanded the scope of GPT-4o's capabilities and enhanced the conversational features to respond more flexibly.
Additionally, to encourage user retention and further engagement with the service, we designed and launched new features.
One of these was a push notification system that alerts users when timely news about their registered favorite stocks is detected.
This allows users to quickly access summarized updates provided by the Stock Agent, enabling them to respond promptly to relevant issues.
Takeaways
Through the process of launching a major feature, I developed critical collaboration skills and holistic thinking. This included foundational tasks such as designing the backend architecture and selecting reliable CP partners to provide trustworthy data.
In particular, the implementation of in-app screens required close collaboration with GUI designers and front-end developers. This experience taught me how to communicate effectively to bring the envisioned output to life, as well as how to support the team as both a collaborator and a project manager.
Looking ahead, I aim to dive deeper into analyzing user behavior and usage patterns to refine the feature. My focus will be on identifying how to make it more accessible to users who need it, addressing areas for improvement with greater attention to detail, and providing a more personalized experience tailored to each user's needs and preferences.