How Arta Finance Is Rebuilding Private Banking from the Inside Out

In Episode 4 of AI-VOLUTION The Series 2026, Sanphat Wangkachonkiat, Principal of Digital Investment at Siam Commercial Bank, sat down with Emmy Sakulrompochai, Product Lead and Global Head of Investment Advisory at Arta Finance, to explore one of the most important questions in wealth management today: can AI automate enough of the advisory process to expand access to quality financial guidance without compromising the standards that regulators and clients expect?

Drawing on eight years at J.P. Morgan and her experience building investment products from the other side of the industry, Emmy shared her perspective on how AI is reshaping investment advisory and where human judgment remains essential.

The Scalability Challenge in Private Banking

Private banking has traditionally been built around highly personalized relationships. Clients work closely with relationship managers who help navigate investment opportunities, market developments, and long-term financial goals. That model can be highly effective, but it is also inherently dependent on human capacity.

As client books grow, relationship managers must balance their time across portfolio reviews, client meetings, regulatory requirements, and administrative work. During periods of market volatility, delivering timely updates and personalized guidance to every client simultaneously becomes increasingly challenging, regardless of the quality of the team involved.

For Emmy Sakulrompochai, who spent eight years at J.P. Morgan across investment banking, equity research, and private banking, these operational realities highlighted an opportunity: could technology handle more of the information gathering, analysis, and routine communication that consumes advisors' time, allowing them to focus on higher-value client conversations?

How Arta Responded: B2C Launch and the B2B Pivot

Arta launched as a B2C platform with a direct proposition: accredited investors and high net worth individuals get access to private banking products through a mobile app or website, with no minimum account size, transparent fees, and a complete product library. KYC is handled online, with no paper forms and no RM cold-calling you to pitch something; information is available whenever you want it.

As the platform matured, banks and wealth managers started approaching them, wanting to know whether the same technology could power their own client-facing services. The B2B white-label model was not in the original roadmap; it emerged from the market. That shift has since become a defining part of how Arta operates: on one side, serving accredited to ultra-high-net-worth investors directly; on the other, deploying agents into bank workflows to handle the manual, time-intensive work that slows financial advisors down.

What Arta Built: Four Capabilities, One Platform

• The Investment Planner uses voice interaction to walk clients through goal-setting and risk profiling, then surfaces a tailored portfolio allocation.
• The Product Specialist handles product-specific questions across the entire product library: fund details, lock-up terms, comparisons between ETFs and structured products. These are the questions that typically stall a decision while someone tracks down the right person to call.
  
• Portfolio Recap delivers on-demand updates: how the portfolio is performing, what happened in the market, how current conditions affect the client's allocation.
• Research Analyst handles open-ended inquiries, where a client reads something interesting and wants exposure. If a client wants to know how to position around gene therapy advances, the capability constructs a tailored investment basket with supporting analytics rather than routing the request to an equity desk. Across all four, the underlying logic is the same: collapse the wait between a client's question and a useful answer.

Under the Hood: Inside the Investment Planner

The Investment Planner is the most technically complex of the four, and Emmy was candid about what sits underneath it: more than twelve AI agents working in sequence. Voice input is translated to text by one agent, a separate agent parses the intent of that text, function calling pulls in real-time data, and the structured output feeds into a proprietary machine learning model that runs convex optimization to produce an asset allocation. Each agent does one job and hands off; the complexity stays underneath.

The design matters for reliability as much as for performance. One agent handling the entire flow would compress steps and carry errors forward. Breaking the workflow into specialist functions means each handoff has a defined input and a defined output; the component that collects a client's goals through voice is not the same component that constructs their portfolio, because those are separate problems solved by separate systems.

Building Compliantly: What It Actually Takes

Arta is registered as a Registered Investment Advisor with the SEC in the United States and works closely with the MAS in Singapore, and that regulatory standing shapes how the agents are designed at a fundamental level. The AI agents do not give investment recommendations. They collect information and pass structured data into the proprietary optimizer; the advice-equivalent output comes from a model with defined, auditable logic rather than from a language model making a judgment call.

Emmy describes the specific design challenge this creates. When a voice agent asks about investment goals and a client responds by asking what they should buy, the agent cannot answer that question; it continues collecting structured inputs until the data is complete, then passes everything to the optimizer. Navigating that constraint naturally, so the interaction does not feel evasive or robotic, required significant iteration. Financial advisors spend time testing the agents, defining ideal responses at every step, and setting hard limits on what the agents cannot say. The guardrails come from domain expertise, not from technical defaults.

Working with regulators as early collaborators, rather than treating compliance as a final gate, has allowed Arta to articulate what it is doing in terms regulators recognize. That relationship with the SEC and MAS is part of what Arta offers banks in the B2B context; the compliance architecture comes with the product.

Why AI Won't Replace Relationship Managers

Emmy is clear that AI is not designed to replace relationship managers. While AI can process information, generate analyses, and respond instantly to routine questions, wealth management often involves conversations that extend beyond portfolio construction and market performance. Clients are making decisions about retirement, family planning, business succession, and long-term financial security—areas where context, judgment, and trust remain essential.

Instead, Arta views AI as a way to augment advisors by reducing the operational work that has traditionally consumed much of their time. Drawing on her experience across investment banking, equity research, and private banking, Emmy observed that advisors often spend significant effort on activities such as onboarding documentation, portfolio analysis, client reporting, and regulatory processes. These tasks are necessary, but they are not always where advisors create the greatest value for clients.

Arta's agents are designed to handle many of these time-intensive workflows. Portfolio updates can be generated automatically, product information can be retrieved on demand, and routine client inquiries can be addressed without requiring manual intervention. By streamlining these processes, advisors can focus more of their attention on strategic discussions, relationship building, and helping clients navigate complex financial decisions.

For Emmy, the opportunity is not simply about efficiency. It is about making high-quality wealth management more scalable. If AI can take on routine operational tasks, advisors can spend more time engaging with clients while maintaining a consistent level of service across a larger client base. In this model, technology does not replace the human relationship at the center of wealth management—it helps strengthen it.

What Wealth Management Looks Like in the Next Three Years

Looking ahead, Emmy expects AI to become increasingly embedded in the day-to-day workflows of wealth management professionals. Rather than operating as standalone products, AI systems are likely to function as co-pilots that help advisors access information more quickly, prepare client materials, and streamline operational tasks.

In this vision, relationship managers remain central to the client experience, but are supported by tools that can surface relevant product information, generate summaries, and automate parts of the reporting and administrative process. The goal is not to replace advisors, but to help them spend more time on client engagement and less time on routine work.

When asked about the growing number of companies building AI solutions for financial services, Emmy pointed to Arta's experience operating both a technology platform and a wealth management business. She argued that building products within a regulated environment provides valuable feedback on how clients interact with advisory services and how compliance requirements shape product design.

Whether AI ultimately transforms access to wealth management at scale remains to be seen. What is clear is that firms across the industry are experimenting with new ways to combine automation, human expertise, and regulatory oversight. Arta's approach offers one perspective on how that future might take shape.

Watch full video at https://youtu.be/yYBLRDV8VEM?si=yqzdylsh2sC1NsvX