Cobo's AI-Native Practices - An interview with COO Lily Z. King

From Individual Productivity to Organizational Transformation: Cobo’s AI Evolution

Over the past year, the impact of AI on enterprises has rapidly evolved from a "personal productivity tool" to a "transformation of organizational capabilities".

For Cobo, this shift arrived earlier and in a more concrete form. While many companies still treat AI as a series of isolated tools, Cobo has begun addressing a deeper question: How can AI truly enter the core operational links of an organization?

We sat down with Lily Z. King, COO of Cobo, to discuss how the firm is embracing AI, the specific role it plays in internal efficiency, and the authentic practices of a company pivoting toward an AI-native future.

Q1: What initiatives has Cobo implemented internally to embrace AI?

Lily Z. King: At Cobo, we did not view AI as merely “equipping employees with an additional tool”; instead, we saw it as an upgrade to the organization’s foundational capabilities. Our progression has been structured across three distinct layers rather than a series of fragmented, bottom-up experiments:

• Layer 1: Infrastructure Ubiquity. We provided every employee with LLM access and ample token quotas. This was not a “perk,” but a move to establish AI access as a fundamental productive force, lowering the threshold for experimentation so everyone can integrate AI into their actual workflows.

• Layer 2: Cognitive and Methodological Development. We conduct continuous internal training. More than just sharing industry trends, we help the team build a shared understanding: AI is not just a chat interface, but a solution for information-dense, cross-system, and cross-team processes that can be structured. We aren’t just training “how to use it,” but “where to place it”.

• Layer 3: Scenario-Based Co-Creation. Recently, we held an internal Agent Co-Creation event. It was inspiring to see that participation wasn’t limited to R&D; colleagues from business, operations, and support roles developed Agents that solve real-world daily problems. This proves that once the infrastructure and methodology are in place, AI innovation permeates the entire business fabric.

Ultimately, AI-native is not about the number of tools purchased or demos created; it is about the organization’s ability to make AI a stable, integral part of business operations. The true divide lies in transitioning from individual efficiency to a complete organizational transformation.

Q2: Do you feel AI is relieving your pressure or forcing you to level up?

Lily Z. King: If I had to choose, I would say it is forcing us to level up. AI does more than increase speed; it rewrites the standard of efficiency itself. In the past, certain tasks could move slowly or rely on manual experience; but when AI compresses information synthesis and preliminary analysis, the organization enters a new rhythm: decisions must be faster, coordination must be tighter, and execution more direct.

This creates two significant changes:

1. Elevated Response Standards: Previously, aligning context across teams required multiple rounds of synchronization. Now, if AI handles the initial context alignment, management naturally expects discussions to be more focused and judgments to be more decisive.

2. Redefining Human Value Density: As information processing becomes automated, a professional’s competitiveness is no longer defined by “organizing materials”. It is defined by the ability to define problems, judge priorities, and make high-quality decisions.

AI is not here to provide a respite; it is pushing the entire organization to a higher operating level.

Q3: Where is the value of AI most immediately apparent?

Lily Z. King: The first wave of value manifests at the information and coordination layers. In any enterprise, the costliest obstacles are often “operational friction”: scattered information, missing context, and elongated communication chains.

AI acts by removing these frictions. It can synthesize context, identify pending items, and highlight risks faster, turning raw data into actionable inputs. For management, the core challenge is often the high cost of preparation—ensuring data is accurate and trade-offs are surfaced. By automating this groundwork, management can focus on what truly matters: strategic direction, resource allocation, and risk control.

AI’s primary value is not replacing judgment, but transforming the preparation chain leading up to it. It replaces friction, not responsibility.

Q4: In one sentence, introduce the Cobo Agent you are most satisfied with.

Lily Z. King: I would point to our internal Agent related to client service quality. While it serves internal operations, it can—within privacy-compliant parameters—analyze client sentiment and context to provide a clear summary of client satisfaction. Previously, this required staff to manually monitor communications and rely on intuition.

I value this Agent not just for the time it saves, but because it converts “subjective perception”—which is often siloed in individuals—into a scalable, reusable organizational capability. It allows subtle but critical signals to be detected and addressed much earlier.

Q5: In an AI-native environment, what is the most critical competency for an employee?

Lily Z. King: Interestingly, the most important skills have become the most “fundamental”: the ability to define problems, the ability to judge results, and the ability to drive delivery. Execution itself is becoming less expensive; the scarcity lies in “knowing what to do and to what extent it is correct”.

• Problem Definition: This goes beyond prompt engineering. It is the ability to deconstruct a task: defining goals, constraints, and priorities. If a problem is defined clearly, AI is a powerful multiplier; if it is vague, AI only accelerates confusion.

• Judging Results: AI generates content rapidly, but speed does not guarantee accuracy. In finance and infrastructure, we cannot delegate the validation of logic, data reliability, or risk exposure to a model. The final quality threshold remains a human responsibility.

• Delivery Ability: AI increases “output density,” but more output does not equal more value. High-value professionals are those who can compress complex information into actionable results that others can execute upon.

Q6: Can you share a moment when you were shocked by AI efficiency?

Lily Z. King: It wasn’t a single task getting faster, but the realization that a process requiring multiple teams and manual context-stitching had been compressed into a single, direct input for judgment. The shock was that the organizational chain had suddenly shortened.

In a company like Cobo—which spans R&D, operations, compliance, and service—the most expensive factor is the cost of passing and re-explaining information. AI acts as an organizational compressor, cutting out redundant coordination and giving time back to judgment and execution.

Q7: What is Cobo’s ultimate objective for its AI transformation?

Lily Z. King: Our goal is for Cobo to become a truly AI-native enterprise. This means AI is systematically integrated into the company’s cycles of perception, judgment, execution, and feedback, forming a new division of labor with humans.

We want to move beyond personal productivity. We envision a future where AI and Agents handle routine tracking, coordination, and feedback loops, allowing our team to focus on high-level judgment, resource optimization, and final decision-making. The competitiveness of a future enterprise will stem from its ability to establish a collaborative mechanism where humans and AI evolve together.

Q8: In a financial-grade environment, what is the prerequisite for using AI?

Lily Z. King: In a financial context, the most important factor is not the power of the model, but whether the boundaries are clear, the process is credible, and the accountability is defined.

Errors in financial infrastructure can lead to asset, compliance, or liability risks. Therefore, AI must be controllable, verifiable, and traceable:

1. Clear Governance: Explicit constraints on what AI can access and execute.

2. Verifiability: The reasoning chain and inputs must be transparent for review.

3. Human Intervention: We must be able to locate, interrupt, and correct issues; AI cannot be an unexplainable black box.

The ultimate challenge in our industry is accountability. We prioritize AI’s ability to operate within a robust governance framework to ensure it provides long-term value to a financial-grade organization.