SubQuery Launches Two Major Products for AI and Blockchain Data

A recap of Hermes and AskSubQuery and how they combine to unlock the future of intelligent Web3 data.

Over the past year, AI has advanced faster than any other technology, and blockchain has quietly become one of the world’s biggest open data sources. Yet combining the two in a practical way has remained a challenge. AI struggles to query blockchain data efficiently, and Web3 apps have no simple way to tap into intelligent agents without massive cost, latency, or engineering overhead.

To solve this, SubQuery has recently announced two major products that work together to unlock AI-powered blockchain experiences at scale:

1. Hermes: a new Bittensor subnet designed to train and incentivize GraphQL AI agents for blockchain data
2. AskSubQuery: a MCP service to enable developers to build AI agents that can query blockchain data easily via GraphQL.

Together, they form a powerful new stack: Hermes is the engine. AskSubQuery is the car.

This article explains what each product does and how they connect to build the future of intelligent Web3 data.

SubQuery Hermes: A Bittensor Subnet 

Hermes was launched as SubQuery’s dedicated Bittensor subnet designed specifically for GraphQL-based blockchain data. Instead of relying on generic, expensive large language models, Hermes trains and evaluates AI agents that understand blockchain schemas, respond quickly, and compose accurate GraphQL queries.

Agents compete, improve, and are rewarded through Bittensor’s incentive mechanisms. This makes Hermes a constantly evolving environment where specialised AI becomes faster, cheaper, and more useful over time.

Hermes is the engine beneath SubQuery’s long-term vision: a network of AI agents that can retrieve, interpret, and act on on-chain data natively and efficiently.

Read the full announcement for more details.

AskSubQuery — Natural Language for Blockchain Data

Shortly before Hermes, SubQuery also launched AskSubQuery, a product that enables AI agent developers to easily integrate a powerful GraphQL Query Agent into their applications.

Instead of writing GraphQL queries or navigating complex schemas, users can simply ask questions like:

“What wallets interacted with this contract last week?”
“Show me today’s NFT mints.”

AskSubQuery exposes this capability via the industry-standard MCP protocol, allowing any AI agent to plug in and instantly support natural-language blockchain queries. It works seamlessly with both SubQuery projects and The Graph subgraphs, making it a truly universal conversational layer for Web3 data.

AskSubQuery turns blockchain exploration from a technical chore into an intuitive conversation.

Read the full announcement for more details.

How They Work Together

The launches were intentionally connected.

• Hermes optimises AI agents to train and understand blockchain data.
• AskSubQuery is an authorized caller which will benefit from Hermes to further optimise the GraphQL agent experience - delivering fast, accurate answers directly to users.

Hermes improves the underlying intelligence; AskSubQuery becomes the accessible interface powered by that intelligence. One is the infrastructure layer, the other is the experience layer. Together, they transform blockchain data from something only developers can work with into something anyone can understand and use.

This combination opens the door to AI-native Web3 applications—intelligent explorers, trading assistants, automated support agents, governance tools, on-chain analytics engines, and more.

Why Bittensor?

Bittensor provides the missing incentive layer required to scale specialised AI for blockchain data. Rather than relying on centralized, general-purpose LLMs, Hermes leverages Bittensor’s decentralized training and evaluation framework to continuously improve GraphQL-native AI agents in the open. On Bittensor, agents are objectively scored on accuracy and latency when querying real on-chain data, and are rewarded directly for performance. This creates a competitive environment where the best blockchain-aware models rise to the top, while weaker or outdated approaches are naturally filtered out. For Hermes, Bittensor isn’t just a deployment platform—it’s the mechanism that ensures AI agents remain fast, verifiable, cost-efficient, and aligned with real developer needs over time

A Step Toward Intelligent Web3

Since 2021, SubQuery has been a leader in blockchain data indexing across more than 300 networks. With Hermes and AskSubQuery, we've extended that leadership into the world of AI—connecting the data layer of Web3 with the intelligence needed to interpret it. 

The reason why is that SubQuery has identified that agentic AI is transforming industries, but Web3 risks missing out. The current approach of using general-purpose LLMs for GraphQL agents creates prohibitive cost and latency, stifling innovation. 

These two launches are more than new products. They’re the foundation of an ecosystem where blockchain data becomes interactive, conversational, and intelligently accessible at scale.

About SubQuery

SubQuery is building the foundational data layer of Web3 — open, scalable, and designed for the AI-driven future.

SubQuery Network provides decentralised data indexers and dRPCs that power thousands of dApps across nearly 300 networks. With AI-assisted tools in the SubQuery SDK and Model Context Protocol (MCP) integration, developers can easily build, deploy, and scale blockchain data infrastructure.SubQuery Hermes (Subnet 82) is a Bittensor subnet that connects AI agents directly to blockchain data through GraphQL. By rewarding developers to create and improve these agents, Hermes enables intelligent on-chain applications.AskSubQuery.xyz provides graphql query MCP as a service. It is also the first product to connect to Hermes Subnet as an Authorized Caller.

With SubQuery, you can unlock intelligence in blockchain data.

​​​​Linktree | Website | Discord | Telegram | Twitter | Blog | Medium | LinkedIn | YouTube