Starting from the end of 2024, Cobo has been exploring the combination of AI and blockchain, in addition to its core cryptocurrency custody and stablecoin payment business.
The earliest potential we identified was the standardization skill potential brought by MCP. Theoretically, if skills are sufficiently standardized, AI can invoke capabilities like a plugin, and blockchain would become the most natural financial infrastructure for AI.
Thus, we incubated an MCP application store internally. But it was soon proven otherwise.
At that time, the threshold for AI was still too high for only seasoned engineers to proficiently invoke it. MCP was not standardized enough, each integration was time-consuming and labor-intensive, costly, slow to advance, and the results were far from expectations.
However, the AI team was built after all. It's expensive, hard to recruit, and cannot be easily removed.
So we decided to change direction. Since we cannot transform the customer world right now, let's start with transforming ourselves.
The First Issue: Security
As an asset custody company, Cobo's data and internal technical process framework are extremely sensitive. There are strict internal data hierarchies. But without data, without real business input, it is impossible to train the company's own Agent.
Initially, we considered local model deployment. But the reality is that the intelligence of local models does not meet the requirements. They can run, but are not usable; they can answer, but are not smart enough.
Ultimately, we chose Claude and Gemini as the main options (we can apply for ZDR—Zero Data Retention Clause to achieve the highest level of isolation).
But large models are merely the underlying "brain" of the business. The truly complex part is data and permissions.
We later created a complete internal knowledge base and Agent framework.
Internal knowledge base + Cobo self-developed agent system
The knowledge base is responsible for the layered structure of the company's internal data. Readable ranges are allocated based on employee permissions.
When Agents invoke the knowledge base, they also inherit employee permissions, rather than having a "God's perspective."
The specifics include:
- How to isolate the network environment
- How to restrict cross-layer data flow
- How to control log retention for auditability
- How to prevent sensitive information leakage
None of these are glamorous, but they determine whether this can sustain in the long run. AI cannot become a security vulnerability.
The Issue After Building the Architecture: No One Uses It
Even today, the company still faces a reality: many frontline businesses are dismissive of AI.
If we only encourage usage, the workflow changes will not happen.
We later realized that we must start with company management.
The First Breakthrough: OKR Agent
The first scenario we strongly promoted was not customer service, nor coding.
It was OKR.
We used AI to break down company strategies, help set OKRs, track progress, and conduct reviews.
In other words, the management of the company slowly shifted from human management to silicon-carbon co-governance. This process has been extremely uncomfortable for employees.
Previously, goals could be written to look pretty, and the process could be explained reasonably. Now, weekly data is available, and excuses are fewer.
From that moment on, goals became not just discussions in meetings, but continuous records in the system.
Strategy OKR weekly urging business progress
But it was also from performance that everyone became truly familiar with AI. Because if you do not participate, it will directly affect your compensation.
From Performance to Business: Comprehensive Agentization
Once OKR started running, we began to promote the agentization of internal services. We enforced each department to establish Agents related to their own business through evaluation + bonuses.
Customer service created customer service Agents. Legal created contract assistant Agents. Sales created CRM Agents.
Finding the most quirky customer agents
In total, we launched over 100 Agents.
We couldn't precisely quantify the results of "silicon-carbon co-governance."
But at least one change is clear:
Previously, when encountering problems, the first reaction was "should we hire another person?". Now the first reaction is, "can we let the system participate first?".
This is actually our understanding of silicon-carbon co-governance. It's not AI replacing people. It's people starting to get used to working with the system.
Some Real Insights from the Journey This Year
First, there is healthy cash flow.
If the company's cash flow is not healthy, this transformation cannot reach the finish line. AI is not a cost-saving tool; it is an upfront investment for long-term structural upgrades, thanks to Cobo's healthy cash flow from its core business.
Second, it must be promoted top-down.
Organizations do not change spontaneously. If the management does not push it strongly, this matter will naturally fail.
As is well known, the founders of Cobo are all heavy AI enthusiasts, and CTO Dr. Jiang has been conducting some AI research since his postdoctoral days at CMU in the early 2000s.
Third, it must be enforced.
If it is only encouraged, AI will forever remain at the level of writing emails. Real changes in processes must carry some "mandatory" aspects.
Fourth, first solve your own business.
Many companies talk about AI + Web3. But if they haven't completed AI transformation internally, all they talk about are concepts.
Looking Back
We also cannot fully quantify this transformation. The company is starting to shift from "human-driven processes" to "goal-driven systems."
If "intelligent organizations" truly emerge in the future, they will certainly not be a result of natural evolution. They will be pushed out through rounds of discomfort.
Because of the participation of everyone, the company can better understand the real needs in the AI era.
This is also a byproduct of our internal transformation.
Recently, we launched Cobo WaaS Skill. Cobo WaaS Skill is an integrated operational capability layer designed specifically for AI Coding Agents, enabling Agents to accurately invoke WaaS APIs through structured knowledge, executable examples, and scenario orchestration. We are upgrading the wallet API to be a financial capability module that can be directly called by AI Agents. The development cycle has been shortened from weeks to conversational levels.
This is not the result of a single product inspiration. Instead, it comes after this round of silicon-carbon co-governance, where capabilities naturally overflow.
We are still exploring.
But at least, today’s Cobo is no longer the same company it was in 2024.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
