At the recent Hubbis Digital Wealth Forum in Singapore, Ada Dong, Vice President of Sales at Eton Solutions, laid out a clear-eyed roadmap for how Artificial Intelligence (AI) and next-generation infrastructure can reshape the operational core of family offices. Drawing on her engineering background and over 15 years of financial industry experience, Dong delivered a compelling call to action: the future of family office efficiency lies not in fragmented tools or short-term AI fixes, but in a unified platform built for privacy, scale, and real-world complexity.
From Excel to AI: The Evolution of the Family Office Operating System
Dong opened by highlighting a structural pain point familiar to many attendees: the proliferation of bespoke, Excel-based workflows and stitched-together point solutions that plague even well-capitalised family offices. “Five years down the road, many family offices realise they’ve been running critical operations off spreadsheets,” she said. “What began as a cost-saving choice becomes a drag on talent, morale, and performance.”
Eton Solutions, founded a decade ago in North Carolina and now operating across 15 countries, offers a different model. Its flagship product, AtlasFive®, is an enterprise resource planning (ERP) system purpose-built for family offices and institutions, enabling seamless performance reporting, accounting, tax ledger, document management, and transaction processing within a single infrastructure.
The firm’s client base spans multi-family offices, private banks, and alternative fund managers, with over 100 families on the platform and a combined USD1.3 trillion of assets under management. In Asia, Eton has established a growing presence across Singapore, Hong Kong, and the Middle East.
The Real AI Gap: It’s Not the Technology, It’s the Integration
Turning to AI, Dong was unflinching in her assessment of industry hype. Despite USD 30 – 40 billion in enterprise AI investments, 95 percent of organisations see little to no return, she noted, citing research from MIT. “The problem isn’t that AI doesn’t work, it’s that organisations aren’t structurally ready to integrate it,” she explained.
According to Dong, three common misconceptions dominate the conversation:
- “Just pick the best Large Language Model (LLM)” – While GPT, Claude, and other models excel in personal productivity tasks, they lack domain specificity, pose privacy risks, and are expensive to fine-tune.
- “Buy a few point AI tools” – Department-level fixes, such as document readers or compliance bots, often lead to inconsistent outcomes across departments and fractured data architectures.
- “Start with sales and marketing” – While appealing due to low sensitivity and fast outcomes, this approach ignores operational inefficiencies in the middle and back office, where the greatest productivity gains lie.
Operationalising AI Without Breaking the System
Dong argued that true transformation comes only when five foundational challenges are addressed:
- Data privacy – AI models must operate within secure, single-tenant environments with no data leakage.
- Data readiness – Documents and transactions must be cleaned, digitised, and mapped into structured formats.
- Workflow alignment – AI must fit existing processes, not demand users adopt new ones.
- Guardrails – Compliance and auditability must be baked in from the outset.
- Contextual accuracy – Retrieval-augmented generation, not hallucinated output, must anchor responses.
EtonAI™, the firm’s proprietary module, attempts to meet these requirements through an architecture that combines International Organization for Standardization (ISO/IEC42001)-certified AI governance, structured document processing, contextual search, and workflow automation. Crucially, all AI processing happens within the client’s chosen Microsoft Azure cloud environment, ensuring that data remains under institutional control.
How EtonAI™Works: From Documents to Decisions
Dong walked the audience through practical applications of EtonAI™ across the front, middle, and back office:
- Intelligent Document Processing: Supports over 250 document types (capital calls, valuations, K-1s, invoices) from day one, using memorised transaction models and patented retrieval tools to auto-log into portals and extract structured data.
- Front Office Agent: Summarises prior client interactions, simulates portfolio impacts, and generates investor tear sheets to assist relationship managers ahead of meetings.
- Middle and Back Office Agent: Automates reconciliation, GL entries, mark-to-market updates, and expense processing, reducing time spent on routine tasks by up to 70 percent and delivering 3 – 4x cost savings.
Use cases range from summarising fund prospectuses against investment mandates, generating meeting minutes from transcripts, drafting tax sheets, to auto-translating PDF reports. “All the long-tail tasks that nobody has time for, AI can now handle them,” Dong remarked.
Why Family Office AI Needs to Be Different
“Generic AI doesn’t understand trust structures, co-ownership layers, or multigenerational governance,” Dong emphasised. As an example, she cited how Eton’s system enables an entity like ‘Jessica’, a co-owner in a complex holding structure, to see only her proportional exposures and permissions, without triggering internal data conflicts.
By integrating fund, asset, and entity-level data with AI-ready formatting, EtonAI™ allows users to trace exposure to a theme (e.g., China biotech) across public ETFs, private funds, and direct holdings, automatically aggregating results in multi-currency views, with or without fees.
Such specificity, Dong argued, is what distinguishes a genuine “Intelligent Family Office Operating Model” from a collection of clever-but-isolated tools. “It’s not just about adding AI to data,” she said. “It’s about building an operating model where AI is native to every task.”
The Compliance Imperative: Certified, Auditable, Trusted
Eton Solutions is one of the few providers globally certified under ISO 42001 for responsible AI management. Additional certifications include ISO 27001 (security), ISO 27701 (privacy), and SOC Type 1 and 2 (reporting integrity). “We don’t treat compliance as an afterthought, it’s built into our foundation,” Dong said.
This is particularly relevant in the wealth management context, where clients expect privacy, regulators demand traceability, and institutions must ensure explainability of AI-driven outcomes.
Client Impact: From Frustration to Productivity
Dong shared that many of Eton’s clients, particularly in Asia and the Middle East, come to the firm after experiencing operational bottlenecks and high staff turnover caused by manual workloads. “People join family offices to do investing, not to reconcile spreadsheets,” she said.
After implementation, clients have reported 80 percent time savings in reconciliation, faster investment research cycles, and measurable improvement in team morale. “When your operations are 90 percent faster, your people spend more time on value creation, not formatting PDFs,” she noted.
Final Thoughts: AI Won’t Replace People, But It Will Replace Inefficiency
Dong concluded with a quote from John G. Taft: “AI will not replace the family office worker… but family office workers working with AI will replace those who don’t.”
It’s a message that resonates deeply in today’s evolving wealth landscape. As the centre of gravity in family offices shifts toward younger generations, digital literacy and operational agility will increasingly define success. For Dong and Eton Solutions, the goal is not to chase trends, but to build infrastructure that lasts.
“The future belongs to intelligent, integrated, and compliant family offices,” she said. “And the time to start building them is now.”
