Moca’s Kenneth Shek: AI Is Having Its MetaMask Moment.

Kenneth Shek has seen this phase of a technology cycle before: impressive infrastructure, excited builders and ordinary users still wondering what to do with it.

Speaking with Blockster’s Ivan Zhelev at TOKEN2049 in Singapore, the Moca Network CEO drew a comparison between today’s AI tools and crypto’s early wallet experience. The first successful transaction could feel extraordinary. Making the technology part of everyday life was another matter.

“I would say right now we're at the MetaMask moment of AI.”

For Shek, the problem is product design. A tool that offers almost unlimited possibilities can still leave its user doing the difficult work of deciding what to ask, connecting services and supervising the result.

His answer starts with narrower, more useful applications. Moca’s part in that effort is identity: helping a business recognize the person behind an automated request, understand what their agent can do and preserve the benefits that person has already earned.

The conversation gives a broader rationale for two developments Blockster has recently covered: Moca Chain’s mainnet launch and AIR’s move into shopping and payments. Both concern what happens when software begins participating in the relationships between customers and businesses.

From Building Models to Building Something People Use

Shek’s route to that view began well before the current chatbot boom. He started his first company while at college in Los Angeles, then moved through solar energy and e-commerce before building a machine-learning business for retailers and consumer companies.

That business developed 30 or 40 models for tasks including supply-chain forecasting and predicting customer behavior. He described working with companies including Marriott Bonvoy and L’Oréal, applying analytics to the practical questions of how customers behave and what businesses need next.

“We also experimented with image generation back then, with our own GPUs, and it was terrible,” he recalled.

The contrast with today is striking. Seven or eight years ago, he said, building AI could mean paying teams to label images before a model could learn what it was looking at. Now, developers and businesses can access sophisticated capabilities through model providers’ APIs.

Shek said he joined Animoca about five years ago. His experience with business data helped steer Moca toward digital identity: the information companies use to understand customers, and the customer’s ability to carry that information elsewhere.

Shek sees a parallel with the proliferation of blockchains. Building infrastructure attracts attention and investment; useful applications still have to follow.

“Most people only want a few things: very simple, easy-to-use and opinionated tools to solve specific problems.”

For Shek, the excitement around new capabilities needs to translate into a clearer purpose.

“What is the AI being used for and how do we make it as simple as possible?” he asked.

That is a practical challenge for companies selling agents. The customer should be able to ask for a result without first learning how to assemble the machinery behind it. Yet the simpler that experience becomes, the more carefully the underlying permissions need to work.

Booking a flight requires more than finding a fare. An agent may need to establish who is traveling, which benefits apply and whether it has permission to complete the purchase.

The Customer Should Have a Say in Where Their Data Goes

Working with consumer businesses also exposed Shek to the less visible side of personalization: information moving between companies.

“Your data is being exchanged behind the scenes amongst businesses without you knowing about it,” he said, describing what he had encountered in business-to-business data sharing.

His objection was both personal and commercial. A company might know a great deal about a customer, while that customer still cannot easily take proof of their status, history or eligibility to another service.

“It’s not scalable because you’re not able to allow the user to prove their own data wherever they go,” he said.

During the conversation, he used Uber and Airbnb as a hypothetical example of the coordination involved in sharing information between businesses. His proposed alternative puts the user at the center: the person retrieves information about themselves and decides where to present it.

The commercial shift would be substantial. A relationship established with one business could become useful to a customer somewhere else, without requiring each pair of companies to build a separate connection.

AI makes that question more urgent in Shek’s view. He envisages agents generating many more interactions than people can manage manually. Those agents would need to carry evidence of the person they represent, together with permission to use it.

That is why Moca’s identity proposition reaches beyond a faster sign-in. It concerns who can put customer data to work, and whose decision makes that use legitimate.

Why Moca Built Its Own Chain

That identity problem also explains Shek’s answer to an obvious question for another Layer 1 project: why build a blockchain at all?

He was blunt about the economics of competing as a general-purpose network. Validators, connectivity and other infrastructure cost money.

“It doesn’t make sense to be the general purpose chain to hopefully compete with Solana and Base and Ethereum to attract liquidity from there,” he said.

Moca’s case rests on a more specific job: maintaining verifiable credentials, supporting encrypted storage and giving applications a common way to check identity and permissions.

Shek said Moca’s roughly one-second finality was fast enough for the job it was designed to do. The priorities he described were verifiable information, encrypted storage, an auditable record and the ability to link an agent to a user’s authority.

That changes what success would look like for the network. A hotel checking a credential or a business withdrawing an outdated claim is meaningful activity even if it produces none of the trading spectacle normally associated with a blockchain launch.

As Blockster reported in “Prove It Once. Use It Again.”, Moca Chain’s mainnet launch brought reusable identity infrastructure together with AIR, the integration suite businesses use for identity, money and loyalty.

Shek illustrated the idea with a hotel checking a guest’s airline status. Instead of accepting an unsupported claim or asking the guest to start the verification process again, the hotel could check a credential from an airline it trusts.

The useful part is the connection to the issuer. Cryptographic verification can establish where a credential came from and whether it remains valid. The business still decides which issuers and claims it accepts.

A reusable credential becomes valuable when it works somewhere beyond the service that issued it. That makes adoption by businesses central to Moca’s proposition.

Why He Welcomes Europe’s Digital Identity Push

Ivan also put the political concern to Shek: digital identity can make services easier to use, but people worry about systems that make them easier to monitor or control.

Shek’s response was unusually positive about Europe’s approach. He sees value in a framework built around people holding credentials and deciding when to share them.

“When you’re able to share data, controlled by the user, it’s actually a good thing,” he said.

He was referring to the revised European digital identity framework, commonly called eIDAS 2.0, and the European Digital Identity Wallet.

The European Commission’s current timetable requires member states to provide wallets by the end of 2026. The requirement is for countries to offer them; using a wallet remains voluntary.

For Shek, clearer rules could initially add friction, then make it easier for businesses to work with identity over time.

“I think EU is actually driving a gold standard for the rest of the world here.”

He said Moca was working toward the European regulatory environment, with more details to share later. That is a statement of direction, rather than an announcement of certification or approval.

His broader point was that privacy and useful data sharing can support one another. A customer who has a clear way to control disclosure may be able to use a credential more widely while revealing less.

Your Identity Should Travel Without Becoming Public

Putting identity infrastructure on a blockchain invites a concern Shek addressed directly: does that make personal information public?

He described a separation between verification records and the underlying information.

“It’s actually stored in a decentralized data storage and all the data is being encrypted with the user’s private key,” he said.

In that model, personal information is not simply published in readable blockchain transactions.

AIR’s mainnet documentation describes cryptographic credential hashes and metadata onchain, with selective disclosure allowing a service to check a relevant claim without collecting all the information behind it.

Shek’s emphasis was control. The person should determine what they present, to whom and for what purpose.

Extending that model to agents creates a second decision. A user may hold a valid credential, but their software still needs a defined right to use it.

Shek made the intended limits concrete with a travel-booking example:

“I want my travel agent A to use my passport information only for 30 seconds, or for a minute, or for an hour.”

“And I can revoke any time after the transaction is done,” he added.

Those examples explain the intended direction. AIR’s launch documentation lists limited, revocable authority for agents as an extension being developed after mainnet, rather than a capability universally available at launch.

They also show why identity and authorization need to remain distinct. Establishing who a customer is does not give an agent unlimited authority to act for them. And withdrawing permission for future access cannot, by itself, erase information a recipient has already received.

The Trade-Off Ivan Pressed Him On

Ivan pushed the discussion beyond the cryptography. If information sits in a storage network outside the user’s device, what happens when the infrastructure holding it fails or is compromised?

Shek’s answer centered on separating the ability to store information from the ability to read and share it. In the model he described, a storage provider holds encrypted data while the user controls the key needed to make that information useful.

He also acknowledged a trade-off. Keeping everything on a person’s device can limit how easily credentials are accessed elsewhere, particularly by software working on that person’s behalf.

“You also actually have to balance the convenience and also the privacy and security of it.”

That balance is central to Moca’s ambition. An agent carrying out a task needs to retrieve the relevant information when the job requires it. The customer still needs control over access.

Encryption addresses one part of that problem. Availability, protection of keys and the design of permission requests still matter. Shek mentioned multifactor authentication and biometrics as ways to add protection, while making clear that he did not want the experience to depend on users managing seed phrases.

There is a product challenge inside that technical answer. People need to understand what they are authorizing without becoming security specialists. A permission that is precise in software but confusing on screen would leave the original usability problem unresolved.

An Internet That Can Verify Without Calling Moca

The longer-term ambition is for verification to work without Moca manually mediating each interaction.

“So without us getting involved as middlemen, the whole internet can be truly autonomous because their data is always available,” Shek said, describing that vision.

He extended the idea across agents verifying other agents, people dealing with businesses and software acting for people. The common requirement is evidence that another participant can check.

That does not remove the need to trust the organization that issued a credential. It gives participants a shared way to check the credential’s origin and status, while the user controls its presentation.

For Moca, that is the bridge between digital identity and the agent economy. Software would be able to establish the authority behind a request as part of completing the task.

The Customer Behind the Bot

The commercial stakes become clearer when Shek returns to travel.

“Expedia knows who I am as a human so that I can apply the right offers, the discounts, the access, the package, the bundles and everything and preventing fraud,” he said.

An automated request arriving from a data center may carry none of that useful customer context. As Shek put it: “You’ve got a million agents’ traffic coming from the same IP address.”

Shek’s concern is that customers could lose the value of those relationships when they delegate shopping to software. The agent might complete a transaction while failing to claim the price, access or benefit its user should receive.

That argument now has a concrete commercial proposal behind it. In Blockster’s October 8 coverage of AIR, Minds, Reap and Pivota, the companies outlined a system connecting verified customer attributes with shopping decisions and controlled payments.

Under the proposed arrangement, AIR would provide identity and eligibility credentials, Minds would coordinate the user’s request, Pivota would help choose and execute the order, and Reap would provide payment credentials limited by merchant, amount and time.

The partnership announcement starts with a controlled demonstration involving one merchant. It is an early test of the model; the announcement does not establish general customer availability.

“You can revoke anytime, you can scope it”

“You can revoke anytime, you can scope it, and you can delegate to your agent to execute too,” Shek said.

His ambition covers the complete shopping journey: “It’s end-to-end, from discovery to transactions to fulfillment, checkout. That can basically be governed by the users and delegated to the agents.”

That is the promise he kept returning to: useful delegation with boundaries the customer controls.

The next test is whether those controls survive an actual purchase. The merchant needs to recognize the credential, the agent needs to stay within its authority, the customer’s benefit needs to apply, and the order needs to complete.

Success would make much of Moca’s infrastructure invisible to the person using it. They would see a booking made correctly, an earned benefit honored and a clear limit on what their agent could do next.

For a founder describing AI’s MetaMask moment, that is an appropriate measure of progress: how much work the technology can take off the customer’s hands.