Since the end of 2024, Cobo—beyond its core businesses of crypto custody and stablecoin payments—has been actively exploring the convergence of AI and blockchain. The first thing we noticed was the standardization potential brought by MCP (Model Context Protocol). In theory, if skills are sufficiently standardized, AI can invoke capabilities like plugins, and blockchain will become AI’s most natural financial infrastructure.
So internally, we incubated an MCP app store. But this idea was quickly disproven. At that time, the AI barrier remained high—only seasoned engineers could skillfully leverage it—and MCP itself wasn’t yet standardized enough. Each integration demanded significant time, effort, and cost, progressed slowly, and delivered far less real-world impact than imagined.
Yet the AI team had already been assembled. It was expensive, hard to recruit for, and couldn’t be easily disbanded. So we pivoted. Since we couldn’t yet reshape our customers’ world, we’d start by reshaping ourselves.
First Challenge: Security
As a digital asset custodian, Cobo handles extremely sensitive data—and its internal technical process frameworks are equally sensitive. We also maintain strict internal data tiering. But without data and without real business inputs, we couldn’t train our own company-specific Agents.
Our initial idea was local model deployment. But reality intervened: locally deployed models simply lacked sufficient intelligence. They ran—but weren’t usable; they answered—but weren’t smart enough. Ultimately, we opted for Claude and Gemini as our primary foundation models (both support ZDR—Zero Data Retention—enabling the highest level of data isolation).
Yet large language models (LLMs) serve only as the underlying “brain” of our business. What’s truly complex is data and permissions. So we built a comprehensive internal knowledge base and Agent framework from scratch.
Internal Knowledge Base + Cobo-Built Agent System
The knowledge base manages internal data tiering. Based on employee permissions, it dynamically assigns read-access scopes. When an Agent queries the knowledge base, it inherits the invoking employee’s permissions—not “god-mode” access.
Key implementation details include:
– How to isolate network environments
– How to restrict cross-tier data flow
– How to control log retention for full auditability
– How to prevent sensitive information leakage
These aren’t glamorous topics—but they determine whether this initiative can run sustainably long-term. AI must never become a security vulnerability.
After the Architecture Was Built: No One Used It
Even today, the company still faces a stark reality: many frontline teams remain dismissive of AI. Merely encouraging usage won’t change workflows. We soon realized we needed to begin at the organizational management layer.
First Breakthrough: OKR Agent
Our first strongly mandated use case wasn’t customer support or code generation—it was OKRs. We used AI to decompose corporate strategy, assist in OKR setting, track progress weekly, and conduct retrospective analyses of bottlenecks.
In other words, we began shifting company management—from human-led management—to silicon-carbon co-governance. This transition was deeply uncomfortable for employees. Previously, goals could be phrased elegantly and processes justified plausibly. Now, weekly data sits openly visible—excuses grow scarce. From that moment onward, goals ceased being just talking points in meetings—they became persistent, system-recorded realities.
Strategy OKR: Weekly Business Progress Tracking
But it was precisely through performance management that everyone truly became familiar with AI—because non-participation directly impacted compensation.
From Performance to Business: Full Agent-ification
Once OKR workflows stabilized, we pushed forward with agent-ifying internal services. Using a combination of benchmarking and performance bonuses, we mandated every department to build Agents aligned with its core functions: Customer Support built a Support Agent; Legal built a Contract Assistance Agent; Sales built a CRM Agent.
Searching for the Most Sarcastic Customer Agent
Ultimately, over 100 Agents went live. We cannot precisely quantify the outcomes of “silicon-carbon co-governance.” But one shift is unmistakable: when problems arise, the instinctive response used to be, “Should we hire another person?” Today, it’s, “Can we first involve the system?”
That, to us, is silicon-carbon co-governance—not AI replacing humans, but humans gradually learning to work alongside systems.
Lessons Learned from This Year’s Journey
First: Healthy cash flow is essential. If your company lacks robust cash flow, such transformation won’t reach completion. AI isn’t a cost-cutting tool—it’s an upfront investment for long-term structural upgrade. We’re grateful that Cobo’s core business continues to generate healthy cash flow.
Second: Top-down enforcement is non-negotiable. Organizations don’t self-transform. Without strong leadership mandate, this initiative would have naturally fizzled out. As widely known, Cobo’s founders are deep AI practitioners—and our CTO, Dr. Jiang, began AI research during his postdoc at CMU in the early 2000s.
Third: Mandatory adoption is critical. If you only encourage usage, AI remains confined to writing emails. Real workflow transformation demands a degree of “enforced adoption.”
Fourth: Start with your own operations. Many companies tout “AI + Web3,” but if you haven’t even AI-ified your internal operations, external claims remain purely conceptual.
Looking Back
We still can’t fully quantify this transformation. Yet the company has gradually shifted—from “people-driven processes” toward “goal-driven systems.” If truly intelligent organizations ever emerge, they won’t evolve organically. They’ll be forged—repeatedly—through rounds of deliberate, uncomfortable pressure.
Because every employee participated, the company gained deeper insight into what’s truly needed in the AI era. That, too, is a valuable byproduct of our internal transformation.
Recently, we launched Cobo WaaS Skill. Cobo WaaS Skill is a dedicated integration and operations layer designed specifically for AI Coding Agents. By providing structured knowledge, executable examples, and scenario orchestration, it enables Agents to accurately invoke WaaS APIs. We’re upgrading wallet APIs into financial capability modules directly callable by AI Agents—shortening development cycles from weeks down to conversational turns.
This wasn’t born from a single product insight. It’s the natural outward spillover of our internal silicon-carbon co-governance journey. We’re still exploring—but at least one thing is certain: today’s Cobo is no longer the same company it was in 2024.
Cobo’s AI Transformation: The Dawn of Silicon-Carbon Co-Governance in Crypto
The crypto industry’s latest strategic pivot toward AI integration received a significant development with Cobo’s detailed account of their internal transformation journey. This isn’t merely another company jumping on the AI bandwagon; it represents a fundamental reimagining of organizational structure within a blockchain-native enterprise, with profound implications for the broader crypto ecosystem.
Strategic Evolution from MCP to Internal Co-Governance
Cobo’s initial attempt at creating an MCP (Model Context Protocol) app store—while theoretically promising—exposed the practical realities of early AI-blockchain integration. Their subsequent pivot to internal transformation demonstrates a sophisticated understanding of technological adoption curves. Rather than forcing external-facing solutions before the technology matured, Cobo wisely chose to build internal capabilities first—a strategy that validates the concept of “eating your own dogfood” in the AI era.
The implementation of “silicon-carbon co-governance” represents perhaps the most significant organizational innovation in crypto management structures since the advent of DAOs. By starting with OKR management before expanding to 100+ internal Agents, Cobo created a flywheel effect where early success in one domain created momentum for broader adoption. This bottom-up approach, enforced through performance management, offers a blueprint for other crypto organizations looking to implement AI without the typical resistance to change.
Market Implications and Competitive Landscape
Cobo’s transformation sends a clear signal to the market: AI is no longer a peripheral experiment but a core competitive differentiator in crypto infrastructure. Their focus on security and permissions—particularly ZDR-enabled models like Claude and Gemini—addresses the industry’s most critical concerns: how to leverage AI without compromising sensitive data.
The launch of Cobo WaaS Skill represents the first tangible product from this internal transformation, signaling the beginning of a phase where accumulated internal capabilities are productized for external consumption. This development creates a new category of AI-blockchain integration services, positioning Cobo as a pioneer in what could become a significant market segment.
For investors, this represents a potential moat development. Companies that successfully implement AI-blockchain integration are likely to achieve operational advantages that competitors without similar capabilities will struggle to match. The fact that Cobo’s core crypto custody and stablecoin businesses funded this transformation suggests a sustainable approach to innovation—one that doesn’t sacrifice current operations for future potential.
Risk Considerations
Despite the optimistic outlook, several risks demand attention:
-
Security Vulnerabilities: The complexity of AI systems introduces new attack vectors. Even with robust permissioning, the risk of prompt injection attacks, data leakage, or model manipulation remains significant for a custody provider like Cobo.
-
Implementation Costs: As Cobo acknowledges, AI transformation requires substantial upfront investment without guaranteed ROI. Their healthy cash flow was essential, but many crypto firms may not have this luxury, potentially widening the gap between well-funded and under-resourced players.
-
Over-Reliance Risks: The shift from “people-driven processes” to “goal-driven systems” could create systemic vulnerabilities. If the AI systems fail or make incorrect decisions, the impact could be magnified across the organization.
-
Talent War: As Cobo noted, recruiting AI talent is expensive and difficult. This creates a potential barrier to entry for smaller firms and could lead to concentration of AI capabilities among well-funded players.
Investment Opportunities
For sophisticated crypto investors, Cobo’s journey reveals several strategic investment themes:
-
AI-Blockchain Infrastructure: Companies building infrastructure that bridges AI and blockchain represent a nascent but rapidly growing category. Look for firms that solve specific pain points in the integration stack, similar to Cobo’s WaaS Skill.
-
Enterprise Crypto Adoption: Firms successfully implementing AI internally are better positioned to serve enterprise clients who will increasingly demand crypto solutions integrated with their own AI systems.
-
Security Innovation: The intersection of AI and blockchain security represents a significant opportunity space. Companies developing novel approaches to AI security in blockchain contexts are well-positioned for growth.
-
Organizational Transformation Services: As more crypto firms recognize the need for AI transformation, service providers who can implement “silicon-carbon co-governance” models will see increasing demand.
Conclusion
Cobo’s transformation journey offers valuable lessons for the entire crypto ecosystem. Their approach—starting with internal operations, implementing mandatory adoption through performance management, and leveraging existing cash flow—provides a pragmatic roadmap for AI integration in blockchain companies.
The concept of “silicon-carbon co-governance” may prove to be as significant for organizational structures in Web3 as DAOs were for decentralized governance. As Cobo demonstrates, the future of successful crypto organizations likely lies not in choosing between human and machine decision-making, but in creating systems where both collaborate effectively.
For investors, the key takeaway is that AI implementation is becoming a critical success factor in crypto, but not all approaches are equal. Those firms that follow Cobo’s pragmatic path—building internally before expanding externally, with careful attention to security and organizational change—will likely emerge as the long-term winners in this new era of AI-blockchain convergence.