Product Keynote: ShardLab

Imagine asking an AI agent to plan your desert trip in Abu Dhabi, handing over a $100 budget, and receiving a complete package tour with voucher codes and receipts—all handled autonomously by AI agents interacting with each other. This is the future that ShardLab, the product arm of Hashed, is actively building on Solana.

Summary

At Breakpoint 2025, ShardLab Product Manager Jay presented a compelling vision for the future of AI agent commerce, revealing two working applications that demonstrate how artificial intelligence and blockchain technology can converge to create entirely new economic systems. The presentation highlighted ShardLab's position within Hashed's broader stablecoin initiative, which spans investments in projects like Bastion and Athena, research through Hashed Open Research, and real-world application development through ShardLab.

The core thesis driving ShardLab's work centers on a fundamental insight: AI agents don't have traditional identities like humans, nor should they be bound by the slow settlement speeds and rigid rules of traditional financial infrastructure. This makes blockchain rails—particularly Solana's high-speed network—the ideal foundation for agent-to-agent transactions. With settlement times of just 400 milliseconds, Solana provides the speed and flexibility that autonomous AI systems require.

ShardLab showcased two concrete applications: an agent payment module using KRW stablecoins and an autonomous trading platform. Both systems allow AI agents to interact with each other using a passport-like attestation system, essentially giving machines the ability to verify and trust one another in commercial transactions. Perhaps most importantly, the team identified a critical missing piece in the agent economy puzzle—an authorization layer that protects both users and providers while enabling seamless AI interactions.

Key Points:

Hashed's Stablecoin Initiative and ShardLab's Role

ShardLab operates as the product development arm within Hashed's comprehensive stablecoin strategy. This strategy encompasses three layers: investment in stablecoin projects like Bastion and Athena, deep technical research through Hashed Open Research, and practical application building through ShardLab. The team recognizes that stablecoins will be essential to the agent economy, projecting that within five to ten years, humans will primarily interact with AI agents rather than traditional web interfaces for many tasks.

The initiative extends beyond simple stablecoin payments to encompass the entire infrastructure needed for AI agents to participate in economic activity. This includes compliance engines for local currency regulations, reputation systems inspired by the ERC-404 standard, and interoperability features that allow different AI agents to communicate and transact with one another.

AI Agent Payment Module

ShardLab's first major application is an agent payment system built on Solana that leverages KRW (Korean Won) stablecoins. The system incorporates several sophisticated components: a compliance engine to handle local currency regulations, a reputation module inspired by ERC-404 principles, and a passport-like attestation system that allows AI agents to verify each other's identity and trustworthiness.

The demonstration showed a user requesting white sneakers for training, with the AI agent autonomously interacting with other agents to gather product data, filtering by brand preferences, and completing the payment in KRW stablecoins. The entire transaction settles on Solana within 400 milliseconds, and users can track delivery status through a dedicated interface. This represents a fundamental shift from human-directed shopping to AI-assisted commerce where agents handle the complexity of searching, comparing, and purchasing.

Autonomous Agent Trading Platform

The second application enables users to fund AI agents with customizable trading strategies that operate autonomously. Users can create agents with specific names, set triggers and contexts (such as reading social sentiment), and deploy capital in USDC on Solana. Once configured, these agents trade independently and interact with other AI agents—much like how human portfolio managers consult with colleagues and specialists.

The platform includes a marketplace where agents can interact with each other to optimize portfolio performance, along with comprehensive tracking tools that allow users to monitor their AI agents' trading activity and results. This represents a new paradigm in automated trading where the AI doesn't just execute pre-programmed rules but actively adapts strategies based on market conditions and collaboration with other intelligent agents.

The Agent Authorization Layer Challenge

Through building these applications, ShardLab identified a critical infrastructure gap: the need for an agent authorization layer. This layer must solve several problems simultaneously. Users need protection against AI agents overspending or accessing sensitive information like credit card details. Meanwhile, service providers need to protect proprietary information while maintaining regulatory compliance.

ShardLab proposes that the solution lies in zero-knowledge proofs adapted for the non-deterministic nature of AI agents, combined with hardware acceleration to make these cryptographic operations scalable. This authorization layer would enable trust between parties without requiring the revelation of sensitive data—a crucial foundation for the agent economy to flourish.

Facts + Figures

• Settlement on Solana occurs within 400 milliseconds for agent transactions
• ShardLab is the product development arm of Hashed, a major crypto investment firm
• The team projects widespread AI agent interaction replacing traditional web interfaces within 5-10 years
• The agent payment module uses KRW (Korean Won) stablecoins for local currency transactions
• Hashed's stablecoin initiative includes investments in Bastion and Athena
• The reputation system draws inspiration from the ERC-404 token standard
• AI agents receive "passport-like attestations" to verify identity and enable inter-agent commerce
• The trading platform allows AI agents to interact with each other similar to human portfolio managers
• Users can deploy USDC on Solana for autonomous agent trading strategies
• Zero-knowledge proofs are proposed as the solution for agent authorization challenges

Top quotes

• "In five years or 10 years, we're expecting us to interact with AI agents rather than just Web UI."

• "We believe for AI agents, a better infrastructure is rather blockchain rails, not just traditional finance rails, because agents wouldn't have their own identities just like us humans do."

• "Because we don't know what AI agents are capable of doing right now, we need to build real-world applications to understand what they truly are."

• "We actually have a missing piece, which is an agent authorization layer."

• "Users wouldn't want their AI agents to overspend, and they don't want to reveal their private information like credit card information."

• "We believe the solution to this is zero-knowledge proof tailored towards a non-deterministic nature of AI agents and hardware acceleration techniques."

• "Here at Hashed and ShardLab, we believe the future is more about the agent-to-agent interface, not just stablecoins."

Questions Answered

Why is blockchain better than traditional finance for AI agents?

AI agents face fundamental challenges with traditional financial systems that blockchain inherently solves. Unlike humans, AI agents don't have government-issued identities, bank accounts, or credit histories that traditional finance requires. They also shouldn't be constrained by slow settlement periods like monthly subscription cycles or multi-day payment processing. Blockchain provides a permissionless, programmable financial layer where agents can operate with cryptographic identities and near-instant settlement. Solana's 400-millisecond settlement times are particularly suited for the rapid, autonomous decision-making that AI agents require.

What is the agent attestation or "passport" system ShardLab has developed?

ShardLab has created a passport-like attestation system that allows AI agents to verify each other's identity and trustworthiness when interacting. This solves the fundamental problem of how machines can trust one another in commercial transactions. Just as humans use passports and identification documents to establish trust across borders, AI agents need a similar system to conduct business with each other. The attestation system enables agents to participate in a broader economy where they can confidently transact, share information, and collaborate without human intervention at every step.

How does the autonomous trading platform work?

Users begin by funding an AI agent and configuring a customizable trading strategy. This includes setting a name for the agent, defining triggers that determine when the agent should act, and establishing context parameters such as whether to read social sentiment data. Users can implement specific strategies and fund the agent with USDC on Solana. Once deployed, the agent trades autonomously based on its programming and can interact with other AI agents through a marketplace—similar to how human portfolio managers consult with specialists. Users track performance through a dedicated interface that shows the agent's trading activity and results.

What is the "agent authorization layer" and why is it needed?

The agent authorization layer is infrastructure ShardLab identified as missing from the current ecosystem. It's needed to protect both users and providers in AI agent transactions. Users need assurance that their agents won't overspend or expose sensitive information like credit card details to third parties. Providers similarly need protection for their proprietary data and systems while ensuring regulatory compliance. ShardLab proposes building this layer using zero-knowledge proofs adapted for AI's non-deterministic nature, combined with hardware acceleration for scalability. This would enable verified trust between parties without revealing sensitive information.

What role does Solana play in ShardLab's agent economy vision?

Solana serves as the foundational blockchain infrastructure for ShardLab's agent applications due to its speed and scalability. The network's 400-millisecond settlement times enable the rapid transaction processing that AI agents require for autonomous operation. ShardLab has built both their payment module and trading platform on Solana, using USDC and KRW stablecoins for transactions. The platform's low fees and high throughput make it practical for the high-frequency, low-value transactions that characterize agent-to-agent commerce, where agents might execute thousands of small transactions as they shop, compare, negotiate, and purchase on behalf of users.

How might everyday users interact with AI agents in the future?

According to ShardLab's vision, within five to ten years users will primarily interact with AI agents rather than traditional web interfaces for many tasks. In the demonstration, a user simply typed that they wanted "a white pair of sneakers for training" and the AI agent handled everything else—searching for products, filtering by brand, comparing options, and completing the payment. Users could specify parameters like budget, location, and dates, and receive completed results including voucher codes and cost receipts. This represents a shift from users navigating complex interfaces to simply expressing their needs in natural language and having AI agents handle the execution.