Master
An AI investing guide representing
a specific style and philosophy
BingX
AI Master
A redesign that makes picking the right Master and
strategy easy, and shows that the AI runs it all
from the first buy to closing the position.
Risk Tolerance
Intuitive TypeBold TypeStrategic TypeLong-term Type
Strategy style
AggressiveModerateConservative
[ 01 — About Service ]
A crypto trading service where you pick a legendary investor’s strategy
and hand the trading over to AI
AI Master is a service where AI trades automatically according to the strategy a user selects.
Users can chat with the AI to get a strategy recommendation, or pick a Master they are interested in and browse the strategies inside it.
The service holds seven Masters, each representing a different investment style, and every Master in turn holds several automated trading strategies.
Strategy
The concrete automated trading
method a Master offers
Once a user picks a Master and a strategy
and commits at least the required minimum,
the AI buys and sells automatically according to that strategy.
In other words, the user decides which strategy of which Master to commit how much to, and the AI carries out the actual trading from there.
[ 02 — Current Experience ]
It was unclear where to start, what to choose,
and how far the AI takes over
The existing screens offered many Masters and strategies, chat recommendations and trading features all at once.
But nothing guided users on what to check first, or in what order to decide.
There was no starting point
Users had to work through seven Masters and theirmany strategies from scratch. Nothing told themwhat to judge a Master by, or what to read first.
What came after a recommendation
Even once a strategy was recommended — by chat orby browsing Masters — it was hard to tell why, howfar the AI trades alone, or what to do next.Even when entering real money, how a strategy ends,how to stop it and how capital settles went unsaid.
Strategy and performance sat apart
Each Master scattered its strategies’ risk levelsand performance, so differences were slow to read.And to see the real price of the coin a strategytraded, users had to leave AI Master entirely.
[ 03.1 — Secondary Research · Community ]
Communities and app reviews alike
said the service was hard to figure out
I looked at how users talked about AI investing on Reddit, Google Reviews and the App Store.
★★★★★
use to difficult not good interface
★★★★★
The interface feels too complex and hard to use. It's difficult to understand what to do.
The reactions in these reviews went beyond “investment strategies are hard.”
Users also struggled to understand where to start in the service, and in what order to use it.
[ 03.2 — Secondary Research · Survey ]
8 of 9 who had used AI features named
“not enough explanation” as the pain point
To see whether the confusion found in the community also showed up in actual use,
I asked 22 crypto exchange app users about their experience with AI-based features.
That included strategy recommendations, copy trading and risk analysis.
59.1%
13 of the 22 respondents had never used an exchange’s AI features. Ten knew about them and chose not to; three did not know they existed.
88.9%
Of the 9 respondents who had actually used AI features, 8 named “not enough explanation” as their biggest pain point.
76.9%
10 of the 13 non-users said they would be willing to try the features if they were improved.
Missing explanation alone cannot account for why people do not use AI features.
But not knowing what the AI does, or how to read its results, looked like one of the things holding them back.
The survey could not tell me at which moment users hesitate, or why they never start trading. So I met users directly and asked about their concrete experience.
[ 04 — Primary Research · Interviews ]
I asked four users of differing experience
why they hesitated over AI trading
I interviewed four users with crypto trading experience, split by how much
they had traded: two beginners and two advanced. I compared how each group
took to AI trading and where they hesitated.
[ 05 — Analysis ]
Beginners struggled to understand it;
advanced users saw no need for AI
Grouping what came up in the interviews by similarity showed that
beginner and advanced participants hesitated for different reasons.
Beginner participants
They did not know what to look at
Beginner participants found it hard to judge which information
mattered, and which features and strategies they actually needed.
In AI Master too, they had no criteria for choosing a Master or a
strategy, so they moved back and forth across several screens.
They did not know how far the AI goes
It was unclear whether the AI only recommends, whether it also
buys and sells, and when a strategy ends.
The doubts grew sharpest on the screen where real money is
committed: how to stop a strategy, and when capital is settled.
Advanced participants
They saw no need to hand things over to AI
Advanced participants already had their own trading principles
and methods. They treated reading the market and judging entry
and exit points as a core part of what makes trading worthwhile.
They would not trust an unproven system
Advanced participants cared about execution speed, system
stability, and the reliability of the trading environment.
To them AI Master read less as a convenient automation tool than
as the risk of handing judgment and capital to an unproven system.
insight
Beginners were interested in AI trading but struggled with how to use it.
Advanced users understood it and still saw little need for it.
So we made beginner users, whose problem clearer explanation should shrink, the primary target.
[ 06 — Competitor Research ]
Other services ask your style and risk first,
then line up return, risk and horizon side by side
Robo-advisors and fund services ask first how you want to invest and what risk you take.
They then show each product’s return, risk and horizon in one format, so differences read easily.
Carmignac · Robo-advisor · Fund Service Interfaces ]
AI Master and robo-advisors serve different products and users.
But in explaining unfamiliar options to a first-time visitor and helping
them choose something that fits, they solve the same problem.
The existing AI Master, by contrast, led with characters and strategies.
Users met a Master’s image and persona before they had established
their own style or compared strategies on shared criteria.
So rather than judging which strategy was closest to them,
users started from names they recognised or images that stood out.
[ 07.1 — Define · Empathy Map ]
Users needed criteria for judging a strategy more than they needed a Master’s image
I pulled what users said and did across the interviews and community research into a single empathy map.
Without clearly understanding the differences between strategies, users took in Masters through impressions like ‘aggressive’ or ‘stable’.
But at the point of committing real money, an image or an impression was not enough to decide on. They needed criteria that connected a strategy to their own situation — the risk they could tolerate, the trading approach they preferred.
[ 07.2 — Define · User Journey ]
There were two paths through the service, and uncertainty spiked on both right before execution
Users could explore strategies in two ways.
- 01.Chatting with the AI to get a strategy recommendation
- 02.Picking a Master they were interested in and browsing its strategies directly
On the chat path, neither the fact that chat was the starting point for recommendations nor how to hold the conversation was made clear.
On the direct path, it was hard to judge which of the many Masters and strategies to choose.
Both paths met at the step of selecting a strategy and entering an amount. At that point users could not tell how far the AI trades automatically, when the strategy ends, or how the capital is settled.
[ 07.3 — Define · HMW ]
From picking a Master and a strategy to committing funds and checking performance,
I framed it as four questions
- 01.
How might we help beginner users choose a Master based on their own investment style?
- 02.
How might we help users understand the current step and how far the AI automates on their behalf?
- 03.
How might we help users check the AI's recommendation in a familiar way and continue the conversation?
- 04.
How might we let users compare strategy performance against real coin prices without leaving the screen?
[ 08.1 — Feature Categorization & Prioritization ]
We kept the ideas most directly connected to the problem
Rather than adding every feature we could, we decided to first solve the problem of users being unable to choose a strategy, and the problem of not understanding what happens after committing money.
We sorted ideas against two criteria.
- Understanding — can users tell where to start, what to choose, and which step they are on?
- Trust — can users tell what the AI does with their money, and when they can step in?
[ 08.2 — Iteration ]
Letting users tune the strategy themselves dissolved the value of handing it to AI
The first idea for building trust was to let users set the buy and sell conditions themselves. We thought that if they could decide how their own money moves, their anxiety would ease.
But that approach put users back in the position of understanding and designing an investment strategy. In the end it also weakened AI Master’s core value — the AI deciding and trading on the user’s behalf.
What beginner participants wanted was not more setup authority, but to know the AI’s judgment and when they could step in. So instead of manual settings, we shifted toward showing who decides at each step and when the user can intervene.
[ 08.3 — Crazy 8s ]
We picked the direction that compares on one screen and shows the next action
Using Crazy 8s — eight ideas sketched in eight minutes — we explored several structures for onboarding, Master recommendation, strategy selection and the execution confirmation screen.
We reviewed each idea against whether the information needed for comparison reads on a single screen, and whether the next action is obvious to the user.
[ 08.4 — Information Architecture ]
Two different ways of exploring, connected into one path to execution
We did not force the way users find a strategy into a single path.
We kept both — one for users who want a recommendation through conversation with the AI, and one for users who want to browse against their own investment style.
Get a recommendation by chatting
- Open chat
- Talk to the AI
- Strategy recommended
Browse it yourself
- Set investment style
- Master recommended
- Browse & pick a strategy
The two paths merge after a strategy is selected.
- Review strategy details
- Confirm capital & automation scope
- Execute strategy
- Track progress
- Exit & settle
Whichever way a user comes in — chat or direct browsing — they confirm what the AI does and where they can intervene in exactly the same way before execution.
[ 08.5 — Wireframe & Design System ]
We validated the core flow as plain screens first, then settled it into one visual system
We first reviewed the order of information and the movement between screens as simple wireframes.
Then we organised components and visual rules so that the current step, the reason for a recommendation, strategy information and the pre-execution notice all read the same way on every screen.
[ 09 — Key Features ]
1. Personalized agent recommendation
[ PROBLEM ]
Users had to work through seven Masters and the many strategies inside them with no clear basis for choosing.
With no criteria to judge by, they fell back on names they recognised or images that stood out.
At the start of the service we ask two things first: the risk they can tolerate, Risk Tolerance, and the trading approach they prefer, Strategy Style.
Once answered, instead of showing every Master at once, we recommend the single Master closest to their investment style. Users go straight to that Master’s screen and start from the strategies it holds.
Cutting the first choice from seven to one lowers the cost of exploring, and lets users look at strategies while already holding criteria tied to their own style.
We did not remove exploration — we gave users a first yardstick for comparing other Masters and strategies against.
2. Current step visibility
[ PROBLEM ]
Users struggled to tell whether they were choosing a Master, selecting a strategy, or actually committing their money.
We added a step structure at the top of the screen showing the current position and the next action.
- Master recommended
- Strategy selected
- Amount confirmed
- Strategy running
Users can see what they are deciding right now, and each screen names the next action concretely — Browse strategies, Check the amount, Start the strategy.
On the final step, where the amount is entered, users re-confirm the AI’s automation scope and how the strategy ends before executing.
As a result, users can locate themselves and their next action without tapping through several screens.
3. Familiar chat interaction
[ PROBLEM ]
AI Master offered strategy recommendations through conversation with the AI. But on the existing screens the shape and placement of the chat icon, and the way a conversation started, differed from familiar messengers — so it never read as the entry point for recommendations.
We changed the shape and placement of the chat icon, and the way a conversation opens, to match the messengers users already know.
Users can open chat and start talking to the AI right away, without learning a new interaction pattern.
That framed chat not as a support feature, but clearly as a path for exploring — talking about the markets and approaches you care about and getting a strategy recommended.
4. Comparison & performance in one flow
[ PROBLEM ]
In the original AI Master you could see a selected strategy’s performance, but the actual price chart of the coin it traded was not connected to it.
To compare performance against real market movement, users had to leave AI Master and find the coin again in the chart list. Moving between screens also broke the context they were building.
We placed the strategy performance chart and the coin’s real price chart on the same screen, swipeable side to side.
- Strategy performance — how the capital moved under this strategy
- Coin price — how the real market price of the traded coin moved
Users can compare a strategy’s performance against real market movement back to back, without leaving AI Master.
Browsing strategy information through to reviewing performance now runs in one continuous flow.
[ 10 — User Test ]
All 5 participants completed the core flow, but strategy details and some icons needed more explanation
Using the redesigned prototype, we tested the core flow: entering an investment style, getting a Master recommended, selecting a strategy, and reviewing the information shown before execution.
All five usability test participants completed the flow from browsing strategies through to the execution confirmation.
The redesign scored an average of 88 on the SUS (System Usability Scale), which rates overall ease of use from 0 to 100 (n=5).
What held up
- All 5 of 5 participants completed the core flow end to end.
- They understood the order — set your investment style first, then get a Master recommended.
- The step indicator at the top let them locate their position and next action.
- On the pre-execution screen they confirmed the scope of automated buying, selling and position exit.
- They checked strategy performance and coin price without leaving the strategy screen.
What to improve next
- The meaning of some feature icons was not obvious on first sight.
- Participants paused to interpret the strategy detail information.
- The order and wording of explanations needs further work so beginners can absorb specialist investment information faster.
With five participants this was an exploratory test, so the SUS and task completion results are read as a reference for where to improve next, not as a settled measure of product performance.
[ 11 — Reflection ]
Letting people know what they handed over mattered more than having AI do more of it
01. In AI UX, an understandable process mattered more than the amount of automation
Users were not asking the AI to simply do more on their behalf.
They wanted to know what the AI was doing with their money, in what order the trading proceeds, and when they could step in.
02. A sense of control is not created by manual settings alone
At first I assumed that giving users more settings would increase their sense of control.
But in an AI service, letting users predict what the AI does and when they can intervene turned out to be just as much a form of control as the range of things they can operate themselves.
03. Trust in AI was tied to how the decision process is explained, not to features
Beyond which Master and strategy the AI recommended, we had to explain why it recommended them, what it handles automatically once trading begins, and when the user can stop it.
Providing the right information at each uncertain moment was the precondition for a user handing money to the AI.
This research was exploratory: 22 survey respondents, 4 interviewees and 5 usability test participants. The interviews in particular had only two participants per group and included experience with comparable AI trading features, so I have not generalised the patterns found into characteristics of the market as a whole.
If this shipped, the first metric I would watch is the activation rate.
Here, activation rate means the share of users who start setting their investment style and go on to select a Master and a strategy and complete their first strategy execution.
That metric would tell me whether clearer recommendation criteria and pre-execution explanation actually convert exploration into a first trade.