Powerpool 2.0: capturing the AI agents market

AI market needs

With new AI programs like Google’s Bard and OpenAI’s ChatGPT, the AI market is set to grow massively. According to the latest stats, it’s predicted to soar from $40 billion, which was reached in 2022, up to $1.3 trillion in the next decade. The growth might happen at a rate of 42% per year, mainly because of the infrastructure for training these AI systems in the short-term period. Later, growth could be noticed in spheres such as language models, digital ads, special software, and services.

Additionally, a growing demand for AI products could bring an extra $280 billion in software sales due to specialized assistants, new infrastructure products, and copilots that make coding faster. Big companies like Amazon WebServices, Microsoft, Google, and Nvidiacan receive the highest benefits as more tasks are transferred to the public cloud.

Source: https://www.bloomberg.com/company/press/generative-ai-to-become-a-1-3-trillion-market-by-2032-research-finds/

The AI segment has received a lot of attention. AI agents supply users with various services, such as generating new data (content) and making decisions based on user requests. To provide such services, AI agents must be pre-trained on existing data.

AI agents automate time-consuming processes that previously required a lot of human intelligence and time. For example, content creation, business decision-making, or scientific applications such as searching for new molecules or drugs in medicine.

AI agents are trained and applied in Web3. They demonstrate promising capabilities in processing off-chain and on-chain data for automatic trading strategies and yield generation. and utilize dApps and protocols more intelligently and efficiently.

The main result provided by AI agents is the decision on specific on-chain actions generated by processing available on-chain data. However, only when the transaction triggered by AI as a result of its operation added to the block is the decision made by the AI agent created on-chain.

In this article, we share an update on PowerPool’s vision, which is focused on providing the backbone for the operation of Web3-focused AI agents. This addition to the existing PowerPool expansion strategy targeting retail users and Defi protocols is a major step forward.

What is PowerPool 2.0?

PowerPool consistently identifies the ongoing challenges in the Web3 market. It offers effective solutions while listening to project needs and the broader community. As the market evolves, PowerPool strives to keep pace.

In 2020, PowerPool was among the first to introduce functioning DeFi indices and meta-governance concepts. Thanks to our DAO and community members, PowerPool evolved into a decentralized network of Keeper-executors for flexible DeFi strategies.

Currently, various networks offer resources for on-chain and off-chain model training. Converting AI-created on-chain actions (triggers) into the updated state according to the AI agent’s decision requires running its Keepers/Bots executing trigger-based transactions.

Considering our experience as a DePIN, we decided to enhance the capabilities of PowerPool Keepers and to streamline the automation of AI agents’ decision execution. The main idea is to utilize PowerPool as an “execution hand” for advanced transaction execution based on AI queries and generated triggers.

PowerPool network as a tool for implementing AI decisions on-chain

PowerPool develops and operates as a DePIN network of Keeper bots. The network’s main service is the automatic execution of on-chain transactions based on on-chain and off-chain triggers.

The service provided by the PowerPool network is Transaction Execution as a Service. It executes transactions and their sequences (referred to as Jobs and flows thereafter) according to the execution algorithms. Applied to the Web3 AI agents sector, PowerPool can be an essential tool supplementing the operation of AI agents on-chain.

Let’s check out theAI agent (Web2 vs Web3) operational cycle:

Web2:

Pre-Training -> Receiving user’s prompt -> generating output -> implementing output and delivering the final result to the user (provide content/data to user or use the API to make some action on the Internet such as sending e-mail).

Web3:

Pre-Training (often on on-chain data) -> Receiving user’s prompt -> generating output (what exactly to do on-chain, e.g., buy token or stake token in a specific contract) -> [!] output implementation and delivering the final result as a state update in the blockchain network (executed transaction on behalf of the user).

AI agents operate off-chain and cannot execute transactions themselves. The output implementation phase mentioned above requires the generation and execution of transactions on-chain. These AI agent protocols must be augmented with a special infrastructure that includes a keeper bot to technically execute transactions on-chain, execution algorithms, RPC on each supported chain, and MEV protection if needed.

There are several options for AI protocol designers to make the output implementation phase possible:

1. Create their own protocol-specific centralized Bots to execute transactions on behalf of AI models launched by users.
2. Build their decentralized infrastructure (a Keeper bot network) with cryptoeconomic security mechanisms and other necessary features.
3. Outsource the transaction execution part of the AI agent cycle to the dedicated DePIN network of Keeper bots, entirely focused on transaction execution

Considering the options above, there are the following comments:

1. Centralized bot infrastructure creates a single point of failure. It exposes the AI protocol to a significant risk of failed executions, leading to user losses and other problems.
2. A decentralized Keeper bot infrastructure is not a core business for the AI agent protocols. However, some early projects have followed this path, creating their own keeper networks for just one protocol. However, it is fundamentally out of scope for most AI protocols since building and running a properly decentralized network on different chains takes a lot of time and effort. It does not make sense to reinvent the wheel when there are networks on the market that are dedicated to solving the problem of automated transaction execution.
3. Outsourcing transactions to existing DePIN Keeper networks and offering Transaction Execution as a Service seems preferable, as it eliminates the need to operate and maintain a dedicated infrastructure. In turn, it allows new AI agent models to be deployed as soon as they are developed, significantly reducing the time required for mainnet on-chain testing and overall go-to-market.

The design of the PowerPool network allows the creation of off-chain triggers that retrieve required data by making API calls and initiating job execution. This unlocks a huge opportunity to execute transactions on behalf of AI agents according to their decision-making strategies.

PowerPool has the following advantages to share with AI agents protocols and their users:

1. It offers transaction execution services with a wide range of conditions (on-demand execution or any kind of strategy) using the AI agents’ protocol API.

It means easy onboarding. So, the protocol that intends to use PowerPool to propagate AI agents’ actions on-chain doesn’t need to develop something specific.

2. It is executed by the decentralized and permissionless network built focusing on robustness and autonomous operation.

It means reliability and simplicity — the Tx execution is secured by PowerPool’s diversified Keepers’ set coordinated by staking/slashing mechanics and not related to the AI agents protocol itself. The AI agents’ protocol doesn’t need to run or maintain any kind of specific infrastructure. It works simply out of the box.

Used for AI transaction execution, PowerPool acts as a DePIN layer for AI agent protocols. By providing reliable and cost-effective propagation of off-chain AI decisions to on-chain state updates, PowerPool enhances its value and user experience.

As stated by Vasily Sumanov, Head of Research at PowerPool:

“Our main goal is to facilitate the practical usage of AI Agents in Web3 by providing reliable infrastructure for AI operations and streamlining the user experience for AI strategy creators.

The expansion of the Web3-focused AI segment, coupled with the essential requirement to execute AI operations on blockchain networks, creates a substantial growth opportunity for PowerPool.

The influence of the new AI-focused vision on PowerPool protocol value

The PowerPool protocol’s ability to create value for users, node runners, and the community is tied to the extent of the market it can capture by executing transactions and strategies on behalf of users, including AI agents. This layer is a critical factor in facilitating user interactions with Web3 protocols. As PowerPool generates and completes transactions, node runners generate more fees, which in turn enhances the overall value of the protocol.

“PowerPool is the ultimate transaction growth layer for networks with low transaction fees, large user bases, and substantial liquidity. It unlocks new use cases on top of automated strategies and AI decision implementation, increasing the number of transactions and users’ convenience.”

The most recent DAO proposal focused on future L2 deployments and strategies to expand the user base on these chains.

By expanding its vision and focus to become an infrastructure layer for Web3-focused AI agent protocols, PowerPool greatly expands the potential number of transaction executions performed by PowerPool keepers.

The key factors of transaction executions within PowerPool:

1. The number of L1/L2 chains where PowerPool is deployed and operating
2. The number of customers on specific L1/L2 chains:

2.1. The number of third-party protocols using PowerPool.

2.2. The number of UI/UX integrated PowerPool Template Jobs. It, in turn, requires integration with UI/UX, Template Job development, and many more.

2.3. The number of AI agents that are used for on-chain actions via PowerPool transaction execution.

3. An incentive multiplicator (points) is applied to a particular L1/L2 chain, UI/UX, or AI agent protocol. The multiplicator means the difference in users’ execution activity in case of points provided for a particular L2/UI/AI vs. no rewards at all.

The template metrics are presented below on the example of DCA (Dollar Cost Average) automated strategy:

1. The number of supported networks;
2. The number of unique DCA Jobs launched by users;
3. The total number of transactions;
4. The total TVL deployed by users to launch DCA Strategies.

AI agents can potentially generate a lot of demand for transaction execution without the need to develop a specific product UI or perform complicated integrations, as they work on top of API calls. Given the projected growth of the AI sector in Web3, there is a huge opportunity to capture a share of this growth as an infrastructure network focused on this sector.

So, the PowerPool becomes a kind of proxy network serving the Web3 AI market:

1. The PowerPool Keepers will capture a share of all gas fees paid by AI agents for their on-chain transactions since they receive an execution fee.
   It makes running a PowerPool Keeper an attractive opportunity since it directly allows one to earn fees from the growing Web3 AI sector. In turn, this leads to the growth of the PowerPool network signers’ set, overall robustness, decentralization, and level of the network crypto economic security (as a result of the greater total size of $CVP stakes).
2. The $CVP token will capture value from multiple AI protocols simultaneously, acting as a kind of a (meta)proxy token for the AI agents market as more Web3 AI agent protocols (or their users) utilize PowerPool. The value captured by $CVP staking increases since it captures the fee stream from all executions for AI-related projects and operations.

AI trading models are a significant subdivision of AI services currently hosting projects in the devnet/testnet/mainnet phases. This sector is evidently necessary for transaction execution. The first protocol we’re considering to demonstrate PowerPool’s automation layer is POND. This protocol emphasizes cluster on-chain analysis, allowing users to create their own AI agents to oversee on-chain trading strategies.

Twitter | Discord | YouTube | Telegram | CMC Community | Debank | Medium