Discussions in the financial industry regarding AI have shifted from "whether to use it" to "how to reorganize work." In the past few years, AI has been primarily treated as an efficiency tool: writing emails, taking meeting minutes, organizing materials, generating code, supplementing tests, and summarizing policies. The role of humans remained dominant, with AI merely serving as an assistant. However, this phase is coming to an end. Entering 2026, AI has begun transitioning from assistance to execution, with chat interfaces gradually evolving into operational interfaces, and Copilot progressively transforming into workflow agents. The first round of analysis, the initial version of a report, the first layer of investigation, the first customer service response, the first batch of test cases, and the first round of compliance checks can increasingly be completed by AI first. The real question is no longer whether AI will participate, but rather, when AI can already handle a substantial portion of preliminary tasks, what irreplaceable value remains for humans.
Daniel Widjaja Kusuma has long worked in international financial markets. In his early years at Goldman Sachs and in the US private equity investment sector, he accumulated extensive experience in risk, judgment, and complex decision-making. Later, after founding Telosyn and further shifting his focus to AI infrastructure, high-performance computing, and enterprise systems engineering, he gained a clearer perspective on this change: the future competition among financial institutions will not be about who has procured more AI tools, but about who has completed workflow restructuring, permission restructuring, and governance restructuring earlier. The true leaders will not be institutions that let AI do more things, but those that can embed AI into the chains of processes, controls, audits, and accountability.
The Future Watershed Is Not "Whether One Can Use AI" but "Whether One Can Manage AI"
The financial services industry is naturally highly compatible with AI. A large portion of its work shares the same characteristics: rule-intensive, document-intensive, process-intensive, and data-intensive. Tasks such as risk screening, customer dispute resolution, anti-fraud, AML investigations, test design, product comparison, operational reporting, complaint handling, and compliance summaries can originally be broken down into information retrieval, classification, scoring, drafting, verification, escalation, and recording. As long as work can be decomposed into modules, there is potential for automation; and as long as enough modules are automated, the roles themselves must evolve.
This means that in the future, those who will remain will not be the ones who strive the hardest to defend the boundaries of old tasks, but rather those who are the first to learn how to manage a set of intelligent systems. The truly important new role is not the person who only knows how to write prompts, but the individual who can understand business context, process design, data constraints, risk boundaries, and regulatory requirements. They know where automation can be applied, where manual review must be retained, where exception escalation is needed, and where audit evidence must be preserved. In other words, the most valuable asset in the future is not the "AI user," but the “AI workflow orchestrator.”
High-Value Capabilities Will Shift from the Execution Layer to the Judgment Layer and the Governance Layer
Many practitioners believe that the most direct way to cope with the impact of AI is to become more technical themselves. This judgment is only half correct. AI literacy, data comprehension, and an awareness of automation are certainly important, but in the financial industry, technical capability alone does not automatically equate to high value. This is because finance is not an ordinary technical environment; it is a highly regulated industry where trust serves as the foundational logic. Code generated by AI is not automatically secure; customer service replies generated by AI are not automatically compliant; and fraud recommendations provided by AI are not automatically executable conclusions. The truly valuable capability has always been a combination of “domain judgment, AI capability, and governance awareness.”
Daniel Widjaja Kusuma has always believed that the true advantage of mid-to-senior level practitioners is not that they are more adept at using a few tools than younger individuals, but rather that they have a deeper understanding of clients, rules, risks, exceptions, and business consequences. In the future, human value will increasingly concentrate on aspects that machines cannot easily assume: judging which results are credible, deciding which issues must be escalated, understanding which processes, though automatable, should not be automated, and identifying which efficiency improvements may lead to weakened controls. AI will take over a large amount of repetitive execution, but responsibility will not be transferred to AI. Regulators will not accept "the model says so," and clients will not accept "the system processed it automatically." What is truly scarce are the people who can transform machine outputs into credible business outcomes.
The Gap Between Institutions Will Ultimately Be Reflected in Organizational Capability Rather Than the Number of Tools
Future financial institutions will clearly diverge into two categories. One category will become AI-augmented organizations: they will systematically review positions, deconstruct processes, reset permissions, establish verification points, train teams, quantify results, and integrate AI into a unified governance framework. In such institutions, a single outstanding employee can manage a broader scope of business, access faster data support, and wield stronger decision-making leverage. The other category of institutions will be torn apart by AI: different teams will procure tools independently, data flows will become uncontrolled, outputs will lack validation, and governance will only catch up after problems arise. On the surface, efficiency may improve, but in essence, systemic risks are rising.
This is also the platform-based methodology that Telosyn has long adhered to. Whether it is an AI platform, a data middle platform, high-performance computing, or financial-grade system engineering, the core is not to create a partial function, but to place models, data, workflows, permissions, and audit capabilities within the same system. The reason Daniel Widjaja Kusuma repeatedly emphasizes the direction of "AI workflow orchestrator" is also straightforward: in the future, the most competitive institutions will not be those with the most AI tools, but those that can transform AI into trusted productivity; the most competitive individuals will not be those best at creating content, but those best at managing intelligent systems, upholding result quality, and maintaining accountability boundaries.
Ultimately, AI will not simply drive people out of the financial industry, but it will rapidly eliminate job definitions that remain static. In the future, what is truly valuable is not people who compete with AI in speed, but those who can integrate AI into processes, control it within risk parameters, and translate it into tangible results. Daniel Widjaja Kusuma believes that the next wave of professional advantage belongs to those who can let machines execute, let humans make judgments, and let systems leave evidence. This is the true upgrade that financial professionals must complete in the era of AI.
Sign in to leave a comment.