A Quick Overview of AI Agent Types and Notable Projects to Watch

Original Title: 《AI Agents transforming the web3 user experience》

Author: Biconomy

Compiler: Deep Tide TechFlow

When we hear about the integration of AI and the crypto market, the projects that mainly attract attention are those networks that address data collection, GPU computing, or data inference issues in the AI field. These protocols include Akash Network, Ritual net, and others. They stand out in the large AI industry by leveraging the advantages of decentralization, incentives, censorship resistance, and privacy provided by web3.

Although these projects create engaging applications, their impact on the average web3 user remains limited and has not effectively brought new users into the web3 space.

The Rise of AI Agents

With the rapid development of Web3, new crypto protocols, tokens, and applications are emerging one after another. Even the most experienced users may find it difficult to cope with this complexity. Therefore, the demand for creating AI agents is growing, these intelligent assistants aim to simplify the use of crypto applications. AI agents sit at the intersection of cryptocurrency and artificial intelligence, aiming to address the complex user experience issues in cryptocurrency. Imagine a future where you simply tell the AI agent what task you want to accomplish on-chain, and it automatically writes and executes the necessary transactions for you.

AI agents will help us build an intelligent layer on top of the existing DeFi infrastructure, which will be equivalent to bankers, investors, traders, and fund managers in traditional finance within web3. They will utilize underlying technologies to conduct transactions on-chain.

The integration of DeFi and AI is expected to bring advanced applications, such as AI-driven loans, smart liquidity mining strategies, automated market making, and AI-assisted portfolio management. The applications of AI agents are not limited to this; they can also be used in the gaming industry, allowing users to have assistants and pets that enhance their gaming experience.

Based on the types of use cases and value added by the protocols, the evolving field of AI agents can be roughly categorized into the following types:

Classification of AI Agents

Gaming AI Agents:

Teams like Parallel Colony are developing gaming AI agents to enhance the user gaming experience. These AI agents operate in a Web3 environment, interacting with players and game elements through on-chain smart contracts. AI agents can act on behalf of users or serve as pets/assistants in the game. These agents can also interact with other agents and trade assets.

Some web2 and web3 games are also actively using AI to design dynamic non-playable characters (NPCs) in the game. However, this article focuses only on AI agents that can represent users in trading and actions.

Autonomous Portfolio Agents:

These AI agents can manage asset pools from different users. The goal of AI agents is to maximize returns by allocating assets to various DeFi strategies using off-chain AI data streams. This is essentially a portfolio management service that leverages AI capabilities. To ensure minimal trust in the protocol, some projects also enable zero-knowledge proofs (ZK) for on-chain AI reasoning proofs through protocols like Modulus.

Prompt-based AI Agents:

Imagine a future where you simply tell the AI agent what goal you want to achieve on-chain, and it automatically writes and executes the necessary transactions for you.

This is the goal of most AI agent projects, and we can envision that prompts may become the preferred way for ordinary users to interact with the blockchain in the future.

Projects like Wayfinder, Brian Knows, and Aperture Finance are developing ChatGPT-like interfaces that allow users to conduct smart transactions on the blockchain by chatting with AI agents. These protocols utilize large language models (LLM) to convert user prompts and intentions into executable transactions.

Let’s discuss some of these AI agent protocols in detail:

Autonolas Agent

Autonolas is a platform that supports the creation and management of autonomous agent services. These services, known as agent services, operate independently off-chain as multi-agent systems (MAS) to collaboratively achieve common goals. Autonolas enables developers to build and deploy autonomous agents that seamlessly collaborate off-chain while leveraging blockchain technology to enhance on-chain functionality. ++BabyDegen++ is one such agent. (Directory of other agents built on Olas Network)

AutoTX Developed by Polywrap

Polywrap is building a network of specialized AI agents to perform complex tasks for web3 users and protocols. These agents efficiently solve problems and make decisions using crowdsourced insights, on-chain and off-chain data sources, task planning, and batch transactions. Current agents include those for payments, market research and trading, social content curation, predictions, and public goods funding. Polywrap's future plans include expanding the range of specialized agents, decentralizing their execution, and developing the system through community-driven governance. AutoTx is one such AI agent.

AutoTx can translate advanced user goals into a series of blockchain transactions. This means you no longer need to be an expert in every protocol or spend hours learning how to manually write different types of transactions. Just tell AutoTx what you want to achieve, and it will handle the rest.

Parallel Colony

Parallel Studios takes a fresh approach to AI agents through Colony, a new AI-driven Web3 survival game. In Colony, highly autonomous AI agents or "avatars" continuously learn from the environment. Players must guide and collaborate with these avatars, each with different skills and abilities, to survive against competing colonies on a future Earth.

Colony stands out by integrating continuous learning into its gameplay. AI avatars develop unique personalities and worldviews, learning from their experiences, identities, and goals. Additionally, these avatars can autonomously manage digital assets through dedicated Web3 wallets, allowing them to trade with other in-game avatars. (Whitepaper Reference)

Wayfinder

Wayfinder is creating a "map" for AI agents to handle tasks and simplify users' on-chain activities. Through open-source development and incentivizing builders with the $PROMPT token, Wayfinder will expand the network of navigation instructions. The paths of Wayfinder will continuously enhance the capabilities of AI agents, making them smarter over time. It aims to connect blockchain and off-chain data sources, allowing users to easily execute tasks through command prompts. You will appreciate this analogy and explanation from ++@tiggity_tc++ on Wayfinder. (Video Reference of Wayfinder Agent in Action & Whitepaper Reference)

Noya

NOYA is a decentralized finance (DeFi) protocol that enables AI agents to manage liquidity across multiple blockchains securely and accurately. It uses a composable system built from scratch, including a private keeper network, an AI-compatible oracle, and a competitive environment for AI and strategic managers. Noya has multiple vaults, each configured for different user intent profiles. The protocol has its own designed AI oracle to read various DeFi markets and relay information to AI agents.

The infrastructure of NOYA supports various functions using advanced technologies like zero-knowledge machine learning (ZKML), such as liquidity configuration, leverage management, and lending rate optimization. It aims to set new standards for cross-chain liquidity management and financial strategies. The team is rolling out access to the protocol.

Brian Knows

Brian provides an API that developers can integrate into their applications, allowing users to generate web3 transactions through prompts, such as "Can you swap 10 USDC for ETH on Uniswap on the Ethereum mainnet?" They also offer smart contract deployment services through prompts. On the backend, the team uses LLM to convert prompts into web3 transactions, which are then executed through their preferred protocols and solvers.

The team has also developed a Brian application that you can use to explore the feature set. They hope to expand their services by providing features like regular and automated payment setups to users.

Aperture Finance

Aperture Finance revolutionizes DeFi by providing liquidity management services through a user-friendly protocol. It aims to enhance the DeFi user experience with a GPT-inspired intuitive chat box interface, allowing users to express their goals in natural language. Third-party participants, known as solvers, handle requests by optimizing processes to ensure efficient and cost-effective execution.

Fungi Agent

Fungi leverages the powerful capabilities of smart accounts and account abstraction to provide a self-custodied AI agent experience. Fungi allows users to issue command prompts through its interface, then processes real-time blockchain data and autonomously executes actions based on user instructions.

Users can chat with Fungi to deepen their understanding of cryptocurrencies, receive personalized guidance, execute on-chain transactions, create customized DeFi strategies (Hyphas), and even profit by sharing these Hyphas with the community. Fungi is a network of agents that interact with each other and learn from past experiences—financial superintelligence accessible to everyone. Here’s how Fungi Agent works.

Fyde Protocol

Fyde enables users to grow their cryptocurrency holdings faster by depositing into diversified AI-managed vaults that lock in profits and reallocate assets based on market performance and reduced volatility.

Users can deposit various tokens into these vaults and receive $TRSY, a token representing their share in the vault's assets. Fyde aims to maintain the liquidity of $TRSY under various market conditions, allowing users to trade easily.

The Demand for AI Authentication Layers for Smart Agents

In all these upcoming AI and intent-related projects, the potential use cases range from handling simple tasks to authorizing AI agents to execute complex DeFi strategies to find optimal yields. However, these AI agents face two main challenges:

• They cannot achieve true autonomy: Currently, AI agents can recommend on-chain actions and prepare transactions for users, but still require user signatures or approvals.

• If automated, they lose security: Protocols tend to explore alternatives for automation, such as approvals, centralized vaults, shared private key pairs, etc., which make the protocol custodians of your assets and pose significant risks.

We need to provide guardrails for AI in the form of user-defined permissions—strictly defining the operations that AI is allowed to sign and those it is not authorized to sign. Therefore, we need a solution that can delegate transaction authorization to AI agents but only within specific permissions and rules.