By Shawn Mars
For much of the history of financial technology, software has played a relatively straightforward role. Professionals make decisions, while digital tools help them access information, organize data, complete transactions, and move through complex workflows more efficiently.
Artificial intelligence is beginning to change that relationship.
As AI becomes embedded in financial and enterprise software, products are increasingly able to interpret information, identify patterns, generate recommendations, prioritize work, and automate parts of professional workflows. Software is no longer functioning only as a passive tool waiting for instructions. In some cases, it is becoming an active participant in how work gets done.
For product designer Shuchen Wang, whose career has spanned financial technology, enterprise software, and AI-powered products at organizations including Goldman Sachs, Evercore,Artifact AI, and Laxis, this shift represents more than the arrival of a new technical capability. It is changing what product designers are responsible for designing.
“In traditional software, we often focus on how users move through a system,” Wang says. “With AI, we also have to think about how the system behaves toward the user, what it should do automatically, and when it should step back.”
Her perspective has developed across different stages of financial and enterprise technology, from designing complex digital financial products to working on AI-native software and, more recently, continuing her product design career within the financial services industry.
That experience has given Wang a close view of a broader shift taking place across enterprise software: the transition from systems that primarily organize and present information to systems that increasingly interpret, recommend, and act.
From Designing Interfaces to Designing Behavior
Traditional product design has often centered on questions of usability. Designers determine how information should be organized, how users navigate a process, which actions should be emphasized, and how complicated tasks can be made easier to understand.
Those questions remain important. AI, however, introduces another layer.
A designer may now need to consider whether a system should generate a recommendation before a user asks for one. The product may need to determine which tasks can be completed automatically, which require approval, and when an unusual situation should be escalated for human review.
The design questions begin to expand.
What should the system do on its own?
What information should it use before making a recommendation?
When should a user be asked to confirm an action?
What happens when the system is uncertain?
How easily can someone correct or override an AI-generated result?
Who remains responsible for the final decision?
These are not simply interface decisions. They are decisions about system behavior.
Wang first encountered a version of this challenge through her work in financial technology. At Goldman Sachs, she worked on complex, data-heavy financial products where professionals needed to interpret large amounts of information and navigate sophisticated workflows efficiently. The design challenge was often not to eliminate complexity, but to structure it in a way that helped users understand what mattered and make informed decisions.
As her work moved further into AI-powered and AI-native products, the nature of the design problem evolved.
“Once the system begins recommending or performing actions, the designer is no longer only organizing information,” she explains. “You are helping define the relationship between the user and the intelligence behind the product.”
That relationship is becoming increasingly important as companies move beyond experimenting with individual AI features and begin integrating intelligence into larger professional workflows.
The Hard Part of Enterprise AI May Be the Workflow
The rapid development of generative AI has made sophisticated technical capabilities increasingly accessible. Companies can now incorporate summarization, extraction, classification, recommendation, and conversational interfaces into products far more easily than they could only a few years ago.
But access to AI does not automatically create a useful product.
A powerful model can still create a poor experience if it appears at the wrong moment, lacks the context required for a professional decision, or creates more work than it removes.
In Wang’s view, this is one of the most important challenges facing enterprise AI.
“The question is no longer just whether the AI can perform a task,” she says. “The question is whether that capability actually fits into the way people work.”
As AI becomes more widely available, Wang believes differentiation may increasingly come from something less visible than the underlying model: how effectively intelligence is integrated into real work.
“The next phase of enterprise AI may be less about who can add the most AI features and more about who can integrate intelligence into workflows in a way that actually improves the outcome,” she says.
Finance Creates a Different Automation Problem
The challenge becomes more complex in financial services and accounting, where not every task carries the same level of risk.
Some activities are repetitive and relatively low risk. Organizing documents, extracting structured information, categorizing records, or summarizing routine data may be well suited to a high degree of automation.
“The question is not simply whether AI can automate something,” Wang says. “It is whether that task should be automated, under what conditions, and where human judgment still adds important value.”
This creates a spectrum rather than a binary choice between manual work and automation.
Product teams must determine how much autonomy the system should have at different stages of a workflow. They may need to design approval mechanisms, escalation paths, exception states, permissions, review processes, and ways for professionals to intervene when circumstances fall outside the expected pattern.
A low-risk task might be completed automatically. A higher-risk task might require a recommendation followed by human review. An unusual case might be routed to someone with additional expertise.
The important design question is not simply how much can be automated. It is how different levels of automation should be distributed across the workflow.
In this environment, workflow design becomes inseparable from AI product design.
The Product Designer as a Workflow Architect
As intelligent systems take on more responsibility, the role of the product designer is also expanding.
Designers have traditionally shaped what users see and how they interact with a product. Increasingly, they may also help shape when the product acts, how it responds to uncertainty, and when control returns to a person.
This requires thinking across a larger system.
An AI-powered workflow may involve business rules, automation, human review, model outputs, permissions, data from multiple sources, and different levels of risk. A successful experience depends on how those pieces work together, not simply on the quality of an individual screen.
Wang describes this evolution as a shift from interface design toward what could be called intelligence orchestration.
“Designers are increasingly shaping not just what users see, but when technology acts, when it waits, and when it hands control back to a person,” she says.
The designer therefore becomes a bridge between technical capability and operational reality.
This does not mean product designers replace engineers, data scientists, or domain experts. Instead, their role increasingly involves helping those disciplines come together around the experience of the person using the system.
What can the technology do?
What does the business require?
What does the user need to understand or decide?
Where should automation begin and end?
Increasingly, those questions must be answered together.
A Career Across Financial Technology and AI
Wang’s own career reflects several stages of the evolution taking place across enterprise technology.
At Goldman Sachs, her work centered on financial products where professionals interacted with large amounts of data and complex workflows. Designing for these environments required an understanding of information hierarchy, user decision-making, and the balance between efficiency and necessary complexity.
She later moved into AI-powered SaaS, where the relationship between user and software began to shift. Products were no longer limited to displaying or organizing information. They could increasingly generate content, interpret inputs, and actively assist users in completing tasks.
At Artifact AI, an AI accounting technology company, that progression became more pronounced. Automation was embedded more deeply into the product experience, and the design challenge extended beyond making software easy to use. It also involved thinking about how users move between automated processing and human review, and how different levels of risk influence the way a workflow should behave.
Wang now continues her product design career at Evercore, bringing together experience in both financial services and AI-driven software. While the specific work differs across organizations and stages of her career, the broader thread has remained consistent: designing technology for professional environments where information, decision-making, and workflow complexity intersect.
Her career path has given her a perspective across different generations of enterprise technology.
Software first digitized existing processes.
Then it automated parts of those processes.
Now AI is beginning to participate more directly in analysis, prioritization, recommendation, and execution.
Each step expands the responsibility of product design.
For Wang, the evolution is not simply from traditional software to AI software. It is a shift in the relationship between people and the systems they use to work.
Beyond “Where Can We Add AI?”
The first generation of AI adoption has often been driven by a simple question: Where can AI be added to an existing product?
Wang argues that the more valuable question is different.
“Instead of asking where we can add AI, I think product teams should ask where intelligence can meaningfully improve the workflow,” she says.
The distinction matters.
An AI feature may be impressive in isolation but have little impact if it solves a problem that is not particularly important to the user. By contrast, a relatively simple automation may create significant value if it removes a repetitive bottleneck, helps professionals identify what deserves attention, or reduces the amount of manual coordination required to complete a task.
This requires product teams to understand the workflow before deciding how much intelligence to introduce.
They need to understand where people spend time, where mistakes or delays occur, which decisions require judgment, what information is needed at different moments, and where an automated action could create new risk.
Only then does the question of AI become useful.
It also requires acknowledging that the best result is not always maximum automation.
In many professional environments, the goal is not to remove people from the workflow entirely. It is to use AI to reduce repetitive work so that people can spend more time on judgment, review, communication, and higher-value decisions.
The Next Competitive Advantage
As AI technology continues to mature, access to powerful models is likely to become less distinctive on its own. Many companies will be able to offer similar underlying capabilities.
What may become more difficult to replicate is a deep understanding of how those capabilities should fit into a specific professional environment.
Financial software must account for business rules, organizational structures, regulatory requirements, risk, and the ways professionals actually make decisions. Successful AI products will need to connect these realities with automation in a way that feels natural rather than disruptive.
This may shift the source of competitive advantage.
The question will not only be which company has access to the most capable technology. It will also be which company understands its users, workflows, and domain well enough to apply that technology effectively.
For product designers, this means the future of the profession may extend far beyond designing interfaces.
It will involve shaping systems in which human judgment, intelligent automation, and business processes operate together.
“The future of AI in financial products is not about putting intelligence everywhere,” Wang says. “It is about putting it in the right places, at the right moments, with the right level of human involvement.”
As financial software evolves from interfaces into intelligent systems, the companies that create the most value may not be those that automate the most. They may be the ones that understand most clearly what should be automated, what should remain human, and how the two can work together.
To explore more of Shuchen Wang’s work and professional experience, visit her portfolio or connect with her on LinkedIn.