Introducing the AI Agent SDK
The world of artificial intelligence is evolving at a breakneck pace, and developers are at the forefront of this revolution. Today, we're thrilled to give you an exclusive first look at something that’s set to redefine how you build, deploy, and interact with AI agents. Introducing our upcoming AI Agent SDK—a powerful toolkit designed to empower developers to create intelligent, context-aware agents with unprecedented ease and functionality.
Check out the open-source ai-agent-sdk repo on GitHub.
Why another AI Agent SDK
You might be familiar with developer tools like ai16z's Eliza and the GOAT SDK (Great Onchain Agent Toolkit). These platforms have set high standards, enabling developers to spin up AI agents with just a few lines of code. These agents can generate text based on prompts and maintain context and memory throughout a session, opening up a myriad of possibilities for applications in customer service, content creation, and beyond.
But we're taking it a step further. While these tools have been instrumental in advancing AI agent capabilities, they lack a critical component for blockchain development: direct access to both onchain and offchain data, including private data. The AI Agent SDK fills this gap by providing built-in blockchain data access, seamless blockchain interactions, and a more robust, scalable solution tailored for the future of decentralized applications. No more cobbling together separate data providers; everything you need is in one powerful package.
Covalent’s vision is to shift AI behavior from being siloed or ephemeral to being persistent, transparent, and verifiable on decentralized systems.
The Importance of Onchain and Offchain Data
As we move toward an increasingly decentralized future, the way transactions are conducted is undergoing a fundamental shift. Agent-to-agent transactions—where autonomous AI agents interact directly with each other—are set to become the norm rather than the exception. In this new paradigm, having direct access to onchain data and offchain isn't just a nice-to-have; it's absolutely essential.
Onchain and offchain data provides a transparent and immutable record of all transactions and smart contract interactions on the blockchain. This data is the lifeblood for AI agents operating in decentralized environments, enabling them to make informed decisions, verify transactions, and interact seamlessly with other agents and decentralized applications in an autonomous fashion.
However, existing AI agent tools often lack integrated access to this critical data, forcing developers to rely on separate data providers or cumbersome APIs. This fragmentation can lead to inefficiencies, increased complexity, and potential security risks.
With the AI Agent SDK, you get built-in access to onchain and soon offchain and private data. By integrating blockchain data directly into the SDK, developers are empowered to create agents that can:
Interact directly with smart contracts and blockchain protocols.
Verify and execute transactions autonomously, without the need for intermediaries.
Access real-time and historical blockchain data, ensuring agents operate with the most up-to-date information.
Facilitate secure and transparent agent-to-agent interactions, paving the way for a new class of decentralized applications.
No more piecing together disparate tools or juggling multiple APIs. With the SDK, everything you need is in one powerful, unified package, simplifying development and accelerating innovation in the AI space. The underlying data source for onchain data is built on the same robust Covalent infrastructure that powers hundreds of consumer and enterprise use-cases today.
What’s Coming Next?
The AI Agent SDK is designed with developers in mind, focusing on simplicity, power, and versatility. Here’s a sneak peek into what you can expect next:
1. Effortless Agent Creation
Creating AI agents shouldn’t be complex. With the AI Agent SDK, you can instantiate a fully functional agent with just a few lines of code. Inspired by developer-friendly toolkits like Eliza and GOAT SDK, this solution abstracts away the intricacies of building and managing AI workflows.
Whether you’re building agents for customer support, data analysis, task automation, or conversational interfaces, the SDK provides a plug-and-play architecture that ensures rapid prototyping and deployment. Agents can seamlessly integrate with APIs, databases, and third-party tools, making it easier than ever to handle real-world tasks.
The SDK also offers pre-built modules and customizable templates that eliminate boilerplate code, allowing developers to focus on fine-tuning their agent's capabilities. By enabling developers to spin up agents quickly and scale them effortlessly, the AI Agent SDK reduces the barriers to building powerful, production-ready AI solutions.
2. Advanced Context and Memory Management
AI agents powered by the SDK go beyond basic, transactional interactions—they remember. By maintaining context and memory across sessions, the SDK allows agents to deliver dynamic, personalized, and highly relevant responses, tailored to user needs.
Agents can seamlessly manage short-term and long-term memory, enabling them to track ongoing conversations, recall previous interactions, and adapt their responses as tasks evolve. This is particularly useful for scenarios like customer support, where agents need to follow multi-turn conversations, or in task management systems, where understanding and retaining complex workflows is essential.
The SDK incorporates efficient state management systems that store memory in a way that is secure, lightweight, and highly scalable. Developers can configure how much context to retain, when to offload memory to external storage, or how to fine-tune memory retention policies for performance optimization.
For instance, an AI agent assisting a developer can "remember" past troubleshooting sessions and proactively suggest solutions based on historical issues. Similarly, agents can provide personalized recommendations to users, such as suggesting products in e-commerce or offering curated content in a media application, all driven by retained insights.
By integrating robust context management and memory capabilities, the SDK ensures that agents deliver interactions that feel natural, intelligent, and progressively more intuitive—enhancing user engagement, satisfaction, and trust.
3. Seamless Blockchain Integration
The AI Agent SDK bridges the gap between AI and blockchains, enabling your agents to interact effortlessly with decentralized systems. With built-in support for querying onchain data, executing smart contracts, and verifying blockchain transactions, your AI agents can become powerful tools for decentralized applications.
Imagine an AI agent that can analyze blockchain activity, monitor wallets, and respond to queries like “What’s the current balance of my wallet?” or “Track the top NFT trades over the last 24 hours.” These integrations empower developers to automate workflows across DeFi, NFT marketplaces, GameFi, RWAs and more.
The SDK will also support interoperability across multiple blockchain networks, ensuring your agents can operate across Ethereum, Layer 2 rollups, and other ecosystems. Through its suite of non-AI methods, developers can leverage blockchain-specific functions—like gas estimation, transaction broadcasting, or historical data retrieval—without needing deep blockchain expertise.
4. Building and Fine-tuning Small/Large Language Models
Whether you need a lightweight model for efficiency or a larger one for advanced reasoning, the SDK will support fine-tuning and customization of language models. Developers can adapt pre-trained models to specific tasks, industries, or data sets, ensuring precision and relevance. Built-in tools allow for efficient training on domain-specific data while maintaining low compute requirements, enabling deployment on devices ranging from edge devices to high-powered servers. This capability empowers developers to create tailored AI solutions that meet unique business needs, balancing performance, accuracy, and resource optimization.
5. Comprehensive Documentation and Support
We believe that great tools are backed by great support. The SDK will come with extensive documentation, tutorials, and a dedicated support team to help you every step of the way.
Stay Tuned and Get Involved!
The future of AI agents is bright, and we’re excited to have you with us on this journey. Whether you’re a seasoned developer or just starting out, the AI Agent SDK is designed to unlock new potentials and simplify the complexities of building intelligent, blockchain-integrated agents.
Connect with Us
Follow us on X, abd LinkedIn to stay connected and be part of our growing community.
Join our developer Telegram chat! https://t.me/CXT_Agent_SDK
Check out the open source ai-agent-sdk repo on GitHub.
