BlockPI’s Account Abstraction Round Table Discussion I invited the core contributors and pioneers in Account Abstraction field to talk around the topic of “Who’s driving the account abstraction development forward”. In this Discussion, Lea from BlockPI and Harry from Moonshot Commons were the co-hosts. Among the rest of the guests, Dror Tirosh is the ERC-4337 author and core developer, and this was the first time for him to join a Twitter Space. The other guests, they are all BlockPI’s partners in bundler service. BlockPI has all of their bundler clients integrated into BlockPI’s network and provide users with the automated option to find the optimal bundler server at any given time, as well as provide them with freedom in choosing between and using multiple bundler services with just one click.

Twitter Space Review: https://twitter.com/i/spaces/1lDxLnzraXYGm

Host & Guests

Lea Li (Host)- CMO of BlockPI

Harry (Host)- Co-Founder & CMO of Moonshot Commons

Albert He — Chief Scientist of BlockPI

Dror Tirosh — ERC-4337 Author & Core Developer

Kristof Gazso- Founder & CEO of Pimlico

Marc — Co-Founder of Candide

Michael Messele — Co-Founder & CEO of Etherspot

Major Discussion Topics

1.Who / what are the major players and driving forces pushing forward the development of the AA industry?

2.How did everyone here initially come up with the idea of becoming a bundler service provider? Does running a bundler require a specialized team? What challenges and difficulties may arise when running a bundler?

3.In the original article of ERC4337, it mentions Bundlers must work with block builders. Must Bundlers from all the chains have to work with block builders? And why?

4.Will the front run happen on Layer 2 chains?

5.What is the current development status of AA on different chains? Which chain is likely to witness the widespread adoption of AA first?

Detailed Discussions

The following texts are the modified version, including some complementary contents, of what the guests said during the Twitter Space, so there will be certain differences between the texts below and the original recording of the Twitter Space.

Short Introduction of Every Guests

Harry: Hi all. My name is Harry. My role at the company is Co-founder and CMO. What Moonshot does is a global Web3 Founder’s Community. We help builders around the globe to learn, build and scale from zero to infinity. We’re also working on a dev education platform, which everyone can stay tuned to, it will be announced later this year. So education and also just helping founders to scale in Web3 spaces is our mission. Thank you. Lea.

Albert: . Good evening, everyone. I am Albert He, the Chief Scientist from BlockPI Network. I got a PhD degree in Aerospace Engineering from Kansas University.

And I jumped into the crypto industry four years ago, then hopped on board the BlockPI team around two years ago, right from the get-go of our infrastructure building adventure.

BlockPI Network is a decentralized multi-chain infrastructure provider, offering one-stop services across more than 33 networks and processing over 20B requests every month. Our RPC service is very popular among all users for its stability, fastest response, and great price performance. By supporting ERC-4337 as a pioneer in this industry, we provide the most diversified Bundler services in the market. If you go to our dashboard or official docs, you will find out that we provide four bundler clients from Stackup, Pimlico, Candide, and Ehterspot. They are also the guests today.

Meanwhile, we also keep optimizing our RPC service for AA. For example, we are running an indexed database for UserOperations of ERC4337. If a user wants to query the UserOperation by its hash using eth_getLogs, our load balancer will recognize this RPC request and dispatch it to the indexed database. The performance then can be hundreds of times better than the original Geth database.

Okay, I think that’s all for the introduction of me and BlockPI Network. Thank you.

Dror Tirosh: Hello. I’m Dror. I’ve been in the blockchain space for 5 or 6 years now. For the past two years, I actually worked with Ethereum on the ERC-4337 for its inception with Vitalik. And before that, we worked on Open GSN, which was just a gas abstraction on Ethereum. It is an open source project, a decentralized one. And I don’t know what else to add. Glad to be here.

Kristof Gazso: Okay, I’m Kristof, Founder and CEO of Pimlico. Pimlico is building paymaster and bundler services, and the new generation of smart accounts. We recently became the largest provider of paymaster and bundler services on the main chains, and we are also one of the people who power BlockPI as well.

Marc: Hey, everyone, I’m Marc from Candide Team. We have been involved with ERC 4337 for a little more than a year now. And we build Candide Wallet and Voltaire Python bundler implementation.

Michael Messele: Hi, thanks for having me. Thanks everyone. I’m Michael Messele, CEO of Etherspot. We’ve been providing account abstracted accounts, not necessarily ERC-4337 through the past three and a half years. Cuz Etherspot derived out of the pillow wallet, which was a smart wallet since around 2018. Now, we have got 4337 compliant wallets that include bundler and paymaster services. And we’ve got a continuation of v1 of account abstraction and v2 which is the 4337. I’ve been actively involved in the blockchain space from 2016, but I knew about blockchain around 2012. So in a nutshell, that is me. Thank you.

1.Who / what are the major players and driving forces pushing forward the development of the AA industry?

Albert: If we have to recognize the major players in this AA industry, for now, I would like to categorize it into three sections, one is decentralized applications, one is infrastructure, and the other one is AA users themselves.

First, the AA Dapps. The AA wallet has the closest connections with the users. These wallets have integrated ERC4337 to enhance the user experience, making it comparable to Web 2.0 applications. In fact, any application that incorporates an AA account system plays a crucial role in the AA ecosystem. While AA wallets have seen significant progress and fulfill the most direct demand, other Dapps may not have reached the same level of extensive adoption.

Second, the infrastructure, including the AA infra such as bundler and paymaster services and AA SDK. BlockPI has integrated the open sourced bundler clients from Candide, Stackup, Etherspot and Pimlico, providing freedom for users to choose among those bundler clients. BlockPI also provides feedback to the Bundler teams to help them better understand some issues, so they can move their product development forward and to build a better infrastructure for AA developers.

Third, users. The users for AA have not yet reached widespread adoption, but there is a growing trend. According to on-chain data from the Account Abstraction Dune Panel, there have been just over one million UserOperations recorded across major chains. This number just is the daily transaction volume on Etheruem, and a large portion is from developer tests. However, there is a significant monthly growth rate, indicating a promising trajectory. It is expected that by the end of this year, there will be a more significant breakthrough in user operations, potentially driving wider adoption of AA.

Kristof Gazso: The AA industry comprises several categories of players driving its advancement. These include bundle providers, paymaster providers like Pimlico, and other notable players such as Biconomy, Stackup, Alchemy, Candide, and Etherspot. These players have actively contributed to the AA ecosystem and played a significant role in its development.

In addition to bundler and paymaster providers, there are also natural smart account providers. These providers offer solutions that enable the creation and management of smart accounts, further enhancing the capabilities of the AA ecosystem.

The modularization of the smart account stack has been instrumental in the emergence of these players. It has provided a foundation upon which players can build and innovate. Smart account providers and bundler paymaster providers offer essential building blocks that enable other players to develop their solutions and services. It is worth noting that the AA industry is dynamic and rapidly evolving. New projects emerge on a daily basis, which is amazing.

2.How did everyone here initially come up with the idea of becoming a bundler service provider? Does running a bundler require a specialized team? What challenges and difficulties may arise when running a bundler?

Marc: Candide does not directly operate bundlers themselves. Instead, they rely on RPC providers like BlockPI to run the bundlers. Since running bundlers requires running a chain node. If one wants to support multiple chains, they would need to run nodes for all the networks. Therefore, Candide focuses on developing bundler software but delegates the responsibility of running the actual bundler software to external parties like BlockPI.

Michael Messele: In the earlier version (v1) of Etherspot’s solution, everything is quite monolithic. Ethersopt deployed various components, such as a relayer service and a wallet, on different chains. The relayer service handled meta transactions and other related functionalities. That entailed the sponsor transaction beat being built in there as well.

However, the approach is more modular now. For instance, if there is a chain X that lacks an account abstraction solution, and if Etherspot is the first one to build AA on the chain, they could deploy the entire AA infrastructure for that chain, including Entrypoint contract, bundler service and the nodes for bundler services.

Additionally, Etherspot is actively involved in the peer-to-peer or mempool sharing of bundlers, collaborating closely with the Ethereum Foundation. This initiative aims to enable the sharing of UserOperation requests among different bundlers. Mempool sharing ensures that the entire ecosystem becomes censorship-resistant and resilient. If you were to remove the Etherspot bundler service, the UserOperations would still be shared across other bundlers without relying solely on the Etherspot bundler.

Etherspot does run bundler services. They have taken the initiative to roll out a bundler service on Layer 2 (L2), which presents its own set of challenges. The exact implementation and technical problems of each L2 solution can vary, leading to variations in deployment timelines.

While there may be delays or complexities in the deployment process, it is crucial to prioritize the stability and functionality of the service to provide a seamless user experience.

Kristof Gazso: I spend a lot of time engaging with various communities and observing the efforts of numerous individuals who are involved in building bundlers and paymasters in-house. The skill set required for building a bundler is distinct from that needed for constructing a smart account.

Bundlers are complex systems with numerous interconnected components. They function as critical infrastructure, demanding high levels of uptime, stability, and scalability. These pieces of software are stateful in nature. Whereas a smart account is a smart contract that can be built and deployed. The scalability of smart accounts is managed by the underlying network.

Many of the teams have had trouble building very scalable, very stable bundlers, and they didn’t want to have to put those engineering resources to actually do that building. So we could really speed up the iteration process of this new generation of smart accounts if we actually provide these bundlers as just a plug-and-play piece of software.

Albert: From the infrastructure point of view, during our research, ERC4337 provides a framework for developers to build upon. And this framework ultimately enables users to have a blockchain experience similar to Web2, even without any knowledge of blockchain technology.

As the research on AA continues, BlockPI anticipates that with the widespread adoption of AA, there will be a substantial increase in RPC demands directed towards the Bundler instead of the original RPC node. It’s important to note that these RPC requests are specific to the Bundler and not the raw RPC of Ethereum clients.

In response to this, the Bundler backend will need to make requests to the original RPC nodes, as mentioned by Marc. This presents a natural advantage for RPC providers like BlockPI, who have globally distributed RPC nodes deployed to handle such demands. Thus, providing Bundler services becomes a natural business opportunity for RPC providers.

Regarding the second point raised by Kristof, I agree that running a Bundler client can be accomplished by individuals with basic computer knowledge. Instructions for open-source Bundler clients can be found, allowing users to follow the steps to set up and run a Bundler.

However, if the intention is to offer Bundler services as a reliable and high-quality service to customers, then a dedicated team is necessary. Providing such a service entails considerations such as SLA (Service Level Agreement) and user experience. For instance, if a wallet team runs a Bundler, they are responsible for their own users. Conversely, if an infrastructure provider like BlockPI runs a Bundler service, they are responsible for their users and business partners. Ensuring a smooth functioning Bundler service requires a dedicated operation team to address any issues promptly and maintain a positive user experience.

During the initial configuration of the Bundler client, BlockPI actually encountered several minor issues. However, after effective communication with the bundle teams, these issues were resolved.

Additionally, BlockPI faces the challenge of monitoring the operational status of multiple open-source Bundler clients and providing customers with an automated option to find the optimal Bundler server at any given time. To accomplish this, it requires a health scoring system to conduct real-time assessments. As a decentralized RPC provider, BlockPI has already developed an existing health scoring system to evaluate RPC nodes. However, due to the varying nature of request processing logic in Bundlers, significant effort has been invested in designing and testing a new scoring system for dispatching UserOperations to those bundlers.

3.In the original article of ERC4337, it mentions Bundlers must work with block builders. Must Bundlers from all the chains have to work with block builders? And why?

Dror Tirosh: When designing ERC4337, a bundler is essentially a block builder. Its main function is to process UserOperations, which are similar to normal transactions, and apply a royalty mechanism based on gas prices. Essentially, this mechanism is copied from normal transactions, inheriting both their advantages and disadvantages.

The bundler also relies on the same protection mechanisms as regular transactions, such as assuming that the network will accept the generated block. However, since the Ethereum protocol itself has not been modified, there is no way to enforce that the bundled transactions will be executed on the blockchain.

Currently, to address this issue, the only way for a bundler to obtain a guarantee that it will not face griefing and ensure that the UserOperations it bundles and sends will be executed is by establishing a connection with the block builder.

Indeed, establishing a connection between the bundler and the block builder provides certain guarantees (so the bundle will not be sent to the Ethereum transaction mempool. Here is an example, the PIP 15 of Polygon, eth_sendRawTransactionConditional) The first guarantee is that the UserOperation bundled by the bundler will be included on the blockchain and will not be front-run by others.). The first guarantee is that the UserOperation bundled by the bundler will be included on the blockchain and will not be front-run by others.

The second guarantee is that there will be no state changes before the bundled UserOperation. The bundle is validated against the current state, and it is expected to be included in that state. If someone manages to put a transaction into the block before the bundled UserOperation, it will result in a change of the state. In the case of a normal block builder, this would create an invalid block that would be dropped by the network. This would result in a loss of potential profit. With account abstraction, the bundle will revert.

If a bundler tries to use the mempool without any protection, there is a risk of front-running. Someone will try to front-run it and make profits. This risk exists even for small gains.

As the market evolves, it becomes evident that bundlers require protection against front-running. This validates the need for mechanisms like flashbot APIs that provide protection. The ERC4337 team is working with different blockchains in order to support these unconditional way of putting UserOps bundles into the system.

Furthermore, there is a need for a proper mempool, which is currently under development in collaboration with the community. A mempool protocol would enable different bundlers to share UserOperations. Some might wonder why a bundler should send its UserOperations to another bundler instead of keeping them to itself.

As long as there are not enough UserOperations, Bundlers should put it itself in the next block. But looking forward when the UserOperations mempool will look like the normal mempool, it will become congested with a high volume of transactions and limited block space, the best thing a bundler will be able to do for himself and for its customers is that any transaction that it can put directly into the next block to send into the public mempool. While another bundler may include these transactions on-chain, it ultimately improves the overall service for everyone involved. Keeping UserOperations to oneself instead of sharing them in the public mempool leads to delays for users. This can lead users to switch to other RPC providers that offer better and more efficient services. As RPC APIs begin to support UserOperations, they will likely experience similar shifts in user preferences.

4.Will the front-run happen on Layer 2 chains?

Marc: Most of the L2s run a single sequencer right now, front-running is not a concern. To address this issue when sequencers are fully decentralized, the first step is to encourage L2s to adopt and implement the eth_sendRawTransactionConditional method. Proposals have been made by Polygon and L2s like Arbitrum, and discussions are ongoing on Optimism as well. And now that’s not enough.

Michael Messele: Arbitrum introduced a proposal called Time Boost. Time Boost allows users to boost their timestamp by 0.5 seconds, effectively accelerating their processing by half a second. This advantage is only available to users who choose to pay for it, giving them an edge over those who do not.

In light of the decentralized sequencer and shared mempool I mentioned earlier, the potential for MEV becomes increasingly significant. It is crucial to establish a mechanism that allows certain transactions to protect themselves in this evolving landscape. With bundlers capable of sharing information among themselves, and considering the influx of millions of transactions, such incidents are likely to become more prevalent. As a result, it becomes essential to address this issue as we move forward, given that I possess knowledge of the transactions handled by specific bundlers.

5.What is the current development status of AA on different chains? Which chain is likely to witness the widespread adoption of AA first?

Dror Tirosh: We do work with specific chains. First of all, at the protocol level, ERC-4337 is by design not modifying the protocol. The only change that we do try to push into the different networks is, as we said, that the ability for bundlers to protect themselves from front run. On Etheruem mainnet, flashbot exists. Bundlers should use minimum flashbot protection.

On other chains, when there is no flashbot, we do need the other mechanisms. One such proposal we have introduced is called “send raw transaction conditional.” This proposal aims to provide bundlers with a certain level of assurance that their transactions will not be reverted on-chain, although the level of guarantee may vary depending on the network and the sequence of operations. This proposal is particularly beneficial for networks like Arbitrum, but we are also working towards implementing it on other chains. I think what will push adoption is the application that uses it. And we don’t control the application. Each application decides which networks it runs for other reasons. So we welcome them all.

Michael Messele: And if I add on this, what we see currently is a lot of layer 2s and layer 1s getting interested in the account abstraction side. And some Dapps and services on those platforms are also showing interest. Some of them do hear the hype and are willing to find out what the hype is about without necessarily understanding it. Some of them have not prioritized it high enough, but you do see a lot of interest. So whichever chain wins necessarily depends on which apps or Dapps and services are busy and more connected with the public. And it doesn’t necessarily mean if a certain Dapp is on chain X, it could not also exist on chain Y. So I think it is a function of how successfully certain Dapp is in making its core product usable to the public. And with account abstraction, we expect Web3 Dapps to feel more like Web2. Hence, those that do have an intuitive user experience (UX) and user interface (UI) will stand to win the next phase of adoption.

Albert: I believe the developers mentioned earlier are deploying AA products on multiple chains, especially on layer 2. Based on our observations over the past few months, the majority of UserOperations have occurred on Arbitrum, OP, and Polygon. Well, even though Ethereum is the birthplace of ERC4337, the high gas fees on Ethereum have led usto believe that those layer 2 chains will be the first chains to witness widespread adoption of AA. But I need to add that we discovered an issue on Polygon. As mentioned earlier, bundler transactions sent on Polygon need to have MEV protection. Unlike layer 2 solutions with native MEV protection through sequencers, Polygon’s MEV ecosystem has been developing at a relatively slower pace. We’re looking for other solutions to protect the bundles’ transactions from front run. I don’t know if the send transaction conditional will eliminate this issue or not. We have not explored this in depth, but I think this is a great opportunity for us to learn this proposal and to see what will happen on Polygon in the bundler ecosystem.

Kristof Gazso: When it comes to the chains that are gaining traction, Pimlico’s focus is not solely on specific chains that we target. Our approach at Pimlico is centered around the idea that there are multiple decentralized ecosystems or worlds emerging, and our goal is to encourage these worlds to adopt smart account-based systems. By doing so, these worlds will naturally choose the chains on which they deploy their systems. We do not try to directly influence or favor specific chains. Instead, our approach is to support the success of these worlds, and if they thrive, they will likely deploy their systems on multiple chains.

Marc: I agree with Dror’s perspective that a significant portion of adoption will occur at the app layer. Currently, most application developers on the EVM primarily test their apps using the Metamask wallet and externally-owned accounts (EOA) without considering the possibility of integrating a contract wallet into their applications. In the near future, the networks or chains that see the highest adoption of account abstraction will be those that offer an enriched version of this functionality. Infinitism team, for instance, is actively working on a proposal to introduce ERC4337 as an enriched version for Layer 2 networks to adopt, while maintaining compatibility with the existing ERC4337 standard. By defaulting to or supporting the use of smart contract wallets, these networks can foster an ecosystem of applications that seamlessly interact with such wallets.

BlockPI Network is a decentralized infrastructure provider, offering one-stop services across 33+ networks and processing over 20B requests every month. Being a leading RPC provider, BlockPI’s RPC service is famous for its rock-solid stability, fastest response rate, and best price performance. By supporting ERC-4337 as a pioneer in this industry, BlockPI also is the only one providing the most diversified Bundler services in the market currently.

Website | Twitter | Telegram | Discord | Medium

About Moonshot Commons

Moonshot Commons is based in Hong Kong, Singapore, and New York, it is a global builder community for Gen-Z engineers in Web 3.

Website | Twitter | Medium

About Ethereum Foundation and ERC-4337 Community

Ethereum Foundation is a non-profit and part of a community of organizations and people working to fund protocol development, grow the ecosystem, and advocate for Ethereum. ERC-4337 Community is a community joined by numerous projects and developers who have made contributions to ERC-4337. ERC-4337 introduces Account Abstraction without any modifications to the core protocol. It also takes usability even one step further by introducing paymasters. Through ERC-4337 Account Abstraction is becoming widespread.

Eth Foundation Website | Eth Foundation Twitter | ERC4337 Community Website | ERC4337 Community Twitter

About StackUp

Stackup’s expertise in ERC-4337 is born out of practical experience and a relentless commitment to service excellence, evidenced by a >99.9% uptime. Choosing Stackup means opting for a collaborator who understands the complexities of account abstraction and provides robust, tailored services.

Website | Twitter | Discord

About Pimlico

Pimlico is an infrastructure platform for building the next generation of Smart Wallets. For developers who are developing an ERC-4337 Smart Wallet, Pimlico provide bundlers, verifying paymasters, ERC-20 paymasters, and much more to help them build a more stable wallet and iterate faster.

Website | Twitter

About Candide

Candide allows users to create a smart account and unleash the true power of Account Abstraction. Candide’s developer tooling leverages the power of every smart wallet seamlessly, eliminating the need for special integrations for individual wallets. With its unified approach, developers can effortlessly tap into the features of any ERC-4337 compatible smart wallet, streamlining development and ensuring a consistent experience for users across the entire ecosystem.

Website | Twitter | Discord

About Etherspot

Etherspot is an Account Abstraction SDK, delivering frictionless Web3 UX. Etherspot SDK represents a multi-chain self-custody Smart Contract Wallet platform that caters to the needs of decentralized application, game, and wallet developers. The platform offers seamless Web3 UX solutions that enable quick onboarding of new users while simplifying the intricate blockchain operations.

Website | Twitter | Telegram | Discord

BlockPI’s Account Abstraction Round Table Discussion I invited the core contributors and pioneers in Account Abstraction field to talk around the topic of “Who’s driving the account abstraction development forward”. In this Discussion, Lea from BlockPI and Harry from Moonshot Commons were the co-hosts. Among the rest of the guests, Dror Tirosh is the ERC-4337 author and core developer, and this was the first time for him to join a Twitter Space. The other guests, they are all BlockPI’s partners in bundler service. BlockPI has all of their bundler clients integrated into BlockPI’s network and provide users with the automated option to find the optimal bundler server at any given time, as well as provide them with freedom in choosing between and using multiple bundler services with just one click.

Twitter Space Review: https://twitter.com/i/spaces/1lDxLnzraXYGm

Host & Guests

Lea Li (Host)- CMO of BlockPI

Harry (Host)- Co-Founder & CMO of Moonshot Commons

Albert He — Chief Scientist of BlockPI

Dror Tirosh — ERC-4337 Author & Core Developer

Kristof Gazso- Founder & CEO of Pimlico

Marc — Co-Founder of Candide

Michael Messele — Co-Founder & CEO of Etherspot

Major Discussion Topics

1.Who / what are the major players and driving forces pushing forward the development of the AA industry?

2.How did everyone here initially come up with the idea of becoming a bundler service provider? Does running a bundler require a specialized team? What challenges and difficulties may arise when running a bundler?

3.In the original article of ERC4337, it mentions Bundlers must work with block builders. Must Bundlers from all the chains have to work with block builders? And why?

4.Will the front run happen on Layer 2 chains?

5.What is the current development status of AA on different chains? Which chain is likely to witness the widespread adoption of AA first?

Detailed Discussions

The following texts are the modified version, including some complementary contents, of what the guests said during the Twitter Space, so there will be certain differences between the texts below and the original recording of the Twitter Space.

Short Introduction of Every Guests

Harry: Hi all. My name is Harry. My role at the company is Co-founder and CMO. What Moonshot does is a global Web3 Founder’s Community. We help builders around the globe to learn, build and scale from zero to infinity. We’re also working on a dev education platform, which everyone can stay tuned to, it will be announced later this year. So education and also just helping founders to scale in Web3 spaces is our mission. Thank you. Lea.

Albert: . Good evening, everyone. I am Albert He, the Chief Scientist from BlockPI Network. I got a PhD degree in Aerospace Engineering from Kansas University.

And I jumped into the crypto industry four years ago, then hopped on board the BlockPI team around two years ago, right from the get-go of our infrastructure building adventure.

BlockPI Network is a decentralized multi-chain infrastructure provider, offering one-stop services across more than 33 networks and processing over 20B requests every month. Our RPC service is very popular among all users for its stability, fastest response, and great price performance. By supporting ERC-4337 as a pioneer in this industry, we provide the most diversified Bundler services in the market. If you go to our dashboard or official docs, you will find out that we provide four bundler clients from Stackup, Pimlico, Candide, and Ehterspot. They are also the guests today.

Meanwhile, we also keep optimizing our RPC service for AA. For example, we are running an indexed database for UserOperations of ERC4337. If a user wants to query the UserOperation by its hash using eth_getLogs, our load balancer will recognize this RPC request and dispatch it to the indexed database. The performance then can be hundreds of times better than the original Geth database.

Okay, I think that’s all for the introduction of me and BlockPI Network. Thank you.

Dror Tirosh: Hello. I’m Dror. I’ve been in the blockchain space for 5 or 6 years now. For the past two years, I actually worked with Ethereum on the ERC-4337 for its inception with Vitalik. And before that, we worked on Open GSN, which was just a gas abstraction on Ethereum. It is an open source project, a decentralized one. And I don’t know what else to add. Glad to be here.

Kristof Gazso: Okay, I’m Kristof, Founder and CEO of Pimlico. Pimlico is building paymaster and bundler services, and the new generation of smart accounts. We recently became the largest provider of paymaster and bundler services on the main chains, and we are also one of the people who power BlockPI as well.

Marc: Hey, everyone, I’m Marc from Candide Team. We have been involved with ERC 4337 for a little more than a year now. And we build Candide Wallet and Voltaire Python bundler implementation.

Michael Messele: Hi, thanks for having me. Thanks everyone. I’m Michael Messele, CEO of Etherspot. We’ve been providing account abstracted accounts, not necessarily ERC-4337 through the past three and a half years. Cuz Etherspot derived out of the pillow wallet, which was a smart wallet since around 2018. Now, we have got 4337 compliant wallets that include bundler and paymaster services. And we’ve got a continuation of v1 of account abstraction and v2 which is the 4337. I’ve been actively involved in the blockchain space from 2016, but I knew about blockchain around 2012. So in a nutshell, that is me. Thank you.

1.Who / what are the major players and driving forces pushing forward the development of the AA industry?

Albert: If we have to recognize the major players in this AA industry, for now, I would like to categorize it into three sections, one is decentralized applications, one is infrastructure, and the other one is AA users themselves.

First, the AA Dapps. The AA wallet has the closest connections with the users. These wallets have integrated ERC4337 to enhance the user experience, making it comparable to Web 2.0 applications. In fact, any application that incorporates an AA account system plays a crucial role in the AA ecosystem. While AA wallets have seen significant progress and fulfill the most direct demand, other Dapps may not have reached the same level of extensive adoption.

Second, the infrastructure, including the AA infra such as bundler and paymaster services and AA SDK. BlockPI has integrated the open sourced bundler clients from Candide, Stackup, Etherspot and Pimlico, providing freedom for users to choose among those bundler clients. BlockPI also provides feedback to the Bundler teams to help them better understand some issues, so they can move their product development forward and to build a better infrastructure for AA developers.

Third, users. The users for AA have not yet reached widespread adoption, but there is a growing trend. According to on-chain data from the Account Abstraction Dune Panel, there have been just over one million UserOperations recorded across major chains. This number just is the daily transaction volume on Etheruem, and a large portion is from developer tests. However, there is a significant monthly growth rate, indicating a promising trajectory. It is expected that by the end of this year, there will be a more significant breakthrough in user operations, potentially driving wider adoption of AA.

Kristof Gazso: The AA industry comprises several categories of players driving its advancement. These include bundle providers, paymaster providers like Pimlico, and other notable players such as Biconomy, Stackup, Alchemy, Candide, and Etherspot. These players have actively contributed to the AA ecosystem and played a significant role in its development.

In addition to bundler and paymaster providers, there are also natural smart account providers. These providers offer solutions that enable the creation and management of smart accounts, further enhancing the capabilities of the AA ecosystem.

The modularization of the smart account stack has been instrumental in the emergence of these players. It has provided a foundation upon which players can build and innovate. Smart account providers and bundler paymaster providers offer essential building blocks that enable other players to develop their solutions and services. It is worth noting that the AA industry is dynamic and rapidly evolving. New projects emerge on a daily basis, which is amazing.

2.How did everyone here initially come up with the idea of becoming a bundler service provider? Does running a bundler require a specialized team? What challenges and difficulties may arise when running a bundler?

Marc: Candide does not directly operate bundlers themselves. Instead, they rely on RPC providers like BlockPI to run the bundlers. Since running bundlers requires running a chain node. If one wants to support multiple chains, they would need to run nodes for all the networks. Therefore, Candide focuses on developing bundler software but delegates the responsibility of running the actual bundler software to external parties like BlockPI.

Michael Messele: In the earlier version (v1) of Etherspot’s solution, everything is quite monolithic. Ethersopt deployed various components, such as a relayer service and a wallet, on different chains. The relayer service handled meta transactions and other related functionalities. That entailed the sponsor transaction beat being built in there as well.

However, the approach is more modular now. For instance, if there is a chain X that lacks an account abstraction solution, and if Etherspot is the first one to build AA on the chain, they could deploy the entire AA infrastructure for that chain, including Entrypoint contract, bundler service and the nodes for bundler services.

Additionally, Etherspot is actively involved in the peer-to-peer or mempool sharing of bundlers, collaborating closely with the Ethereum Foundation. This initiative aims to enable the sharing of UserOperation requests among different bundlers. Mempool sharing ensures that the entire ecosystem becomes censorship-resistant and resilient. If you were to remove the Etherspot bundler service, the UserOperations would still be shared across other bundlers without relying solely on the Etherspot bundler.

Etherspot does run bundler services. They have taken the initiative to roll out a bundler service on Layer 2 (L2), which presents its own set of challenges. The exact implementation and technical problems of each L2 solution can vary, leading to variations in deployment timelines.

While there may be delays or complexities in the deployment process, it is crucial to prioritize the stability and functionality of the service to provide a seamless user experience.

Kristof Gazso: I spend a lot of time engaging with various communities and observing the efforts of numerous individuals who are involved in building bundlers and paymasters in-house. The skill set required for building a bundler is distinct from that needed for constructing a smart account.

Bundlers are complex systems with numerous interconnected components. They function as critical infrastructure, demanding high levels of uptime, stability, and scalability. These pieces of software are stateful in nature. Whereas a smart account is a smart contract that can be built and deployed. The scalability of smart accounts is managed by the underlying network.

Many of the teams have had trouble building very scalable, very stable bundlers, and they didn’t want to have to put those engineering resources to actually do that building. So we could really speed up the iteration process of this new generation of smart accounts if we actually provide these bundlers as just a plug-and-play piece of software.

Albert: From the infrastructure point of view, during our research, ERC4337 provides a framework for developers to build upon. And this framework ultimately enables users to have a blockchain experience similar to Web2, even without any knowledge of blockchain technology.

As the research on AA continues, BlockPI anticipates that with the widespread adoption of AA, there will be a substantial increase in RPC demands directed towards the Bundler instead of the original RPC node. It’s important to note that these RPC requests are specific to the Bundler and not the raw RPC of Ethereum clients.

In response to this, the Bundler backend will need to make requests to the original RPC nodes, as mentioned by Marc. This presents a natural advantage for RPC providers like BlockPI, who have globally distributed RPC nodes deployed to handle such demands. Thus, providing Bundler services becomes a natural business opportunity for RPC providers.

Regarding the second point raised by Kristof, I agree that running a Bundler client can be accomplished by individuals with basic computer knowledge. Instructions for open-source Bundler clients can be found, allowing users to follow the steps to set up and run a Bundler.

However, if the intention is to offer Bundler services as a reliable and high-quality service to customers, then a dedicated team is necessary. Providing such a service entails considerations such as SLA (Service Level Agreement) and user experience. For instance, if a wallet team runs a Bundler, they are responsible for their own users. Conversely, if an infrastructure provider like BlockPI runs a Bundler service, they are responsible for their users and business partners. Ensuring a smooth functioning Bundler service requires a dedicated operation team to address any issues promptly and maintain a positive user experience.

During the initial configuration of the Bundler client, BlockPI actually encountered several minor issues. However, after effective communication with the bundle teams, these issues were resolved.

Additionally, BlockPI faces the challenge of monitoring the operational status of multiple open-source Bundler clients and providing customers with an automated option to find the optimal Bundler server at any given time. To accomplish this, it requires a health scoring system to conduct real-time assessments. As a decentralized RPC provider, BlockPI has already developed an existing health scoring system to evaluate RPC nodes. However, due to the varying nature of request processing logic in Bundlers, significant effort has been invested in designing and testing a new scoring system for dispatching UserOperations to those bundlers.

3.In the original article of ERC4337, it mentions Bundlers must work with block builders. Must Bundlers from all the chains have to work with block builders? And why?

Dror Tirosh: When designing ERC4337, a bundler is essentially a block builder. Its main function is to process UserOperations, which are similar to normal transactions, and apply a royalty mechanism based on gas prices. Essentially, this mechanism is copied from normal transactions, inheriting both their advantages and disadvantages.

The bundler also relies on the same protection mechanisms as regular transactions, such as assuming that the network will accept the generated block. However, since the Ethereum protocol itself has not been modified, there is no way to enforce that the bundled transactions will be executed on the blockchain.

Currently, to address this issue, the only way for a bundler to obtain a guarantee that it will not face griefing and ensure that the UserOperations it bundles and sends will be executed is by establishing a connection with the block builder.

Indeed, establishing a connection between the bundler and the block builder provides certain guarantees (so the bundle will not be sent to the Ethereum transaction mempool. Here is an example, the PIP 15 of Polygon, eth_sendRawTransactionConditional) The first guarantee is that the UserOperation bundled by the bundler will be included on the blockchain and will not be front-run by others.). The first guarantee is that the UserOperation bundled by the bundler will be included on the blockchain and will not be front-run by others.

The second guarantee is that there will be no state changes before the bundled UserOperation. The bundle is validated against the current state, and it is expected to be included in that state. If someone manages to put a transaction into the block before the bundled UserOperation, it will result in a change of the state. In the case of a normal block builder, this would create an invalid block that would be dropped by the network. This would result in a loss of potential profit. With account abstraction, the bundle will revert.

If a bundler tries to use the mempool without any protection, there is a risk of front-running. Someone will try to front-run it and make profits. This risk exists even for small gains.

As the market evolves, it becomes evident that bundlers require protection against front-running. This validates the need for mechanisms like flashbot APIs that provide protection. The ERC4337 team is working with different blockchains in order to support these unconditional way of putting UserOps bundles into the system.

Furthermore, there is a need for a proper mempool, which is currently under development in collaboration with the community. A mempool protocol would enable different bundlers to share UserOperations. Some might wonder why a bundler should send its UserOperations to another bundler instead of keeping them to itself.

As long as there are not enough UserOperations, Bundlers should put it itself in the next block. But looking forward when the UserOperations mempool will look like the normal mempool, it will become congested with a high volume of transactions and limited block space, the best thing a bundler will be able to do for himself and for its customers is that any transaction that it can put directly into the next block to send into the public mempool. While another bundler may include these transactions on-chain, it ultimately improves the overall service for everyone involved. Keeping UserOperations to oneself instead of sharing them in the public mempool leads to delays for users. This can lead users to switch to other RPC providers that offer better and more efficient services. As RPC APIs begin to support UserOperations, they will likely experience similar shifts in user preferences.

4.Will the front-run happen on Layer 2 chains?

Marc: Most of the L2s run a single sequencer right now, front-running is not a concern. To address this issue when sequencers are fully decentralized, the first step is to encourage L2s to adopt and implement the eth_sendRawTransactionConditional method. Proposals have been made by Polygon and L2s like Arbitrum, and discussions are ongoing on Optimism as well. And now that’s not enough.

Michael Messele: Arbitrum introduced a proposal called Time Boost. Time Boost allows users to boost their timestamp by 0.5 seconds, effectively accelerating their processing by half a second. This advantage is only available to users who choose to pay for it, giving them an edge over those who do not.

In light of the decentralized sequencer and shared mempool I mentioned earlier, the potential for MEV becomes increasingly significant. It is crucial to establish a mechanism that allows certain transactions to protect themselves in this evolving landscape. With bundlers capable of sharing information among themselves, and considering the influx of millions of transactions, such incidents are likely to become more prevalent. As a result, it becomes essential to address this issue as we move forward, given that I possess knowledge of the transactions handled by specific bundlers.

5.What is the current development status of AA on different chains? Which chain is likely to witness the widespread adoption of AA first?

Dror Tirosh: We do work with specific chains. First of all, at the protocol level, ERC-4337 is by design not modifying the protocol. The only change that we do try to push into the different networks is, as we said, that the ability for bundlers to protect themselves from front run. On Etheruem mainnet, flashbot exists. Bundlers should use minimum flashbot protection.

On other chains, when there is no flashbot, we do need the other mechanisms. One such proposal we have introduced is called “send raw transaction conditional.” This proposal aims to provide bundlers with a certain level of assurance that their transactions will not be reverted on-chain, although the level of guarantee may vary depending on the network and the sequence of operations. This proposal is particularly beneficial for networks like Arbitrum, but we are also working towards implementing it on other chains. I think what will push adoption is the application that uses it. And we don’t control the application. Each application decides which networks it runs for other reasons. So we welcome them all.

Michael Messele: And if I add on this, what we see currently is a lot of layer 2s and layer 1s getting interested in the account abstraction side. And some Dapps and services on those platforms are also showing interest. Some of them do hear the hype and are willing to find out what the hype is about without necessarily understanding it. Some of them have not prioritized it high enough, but you do see a lot of interest. So whichever chain wins necessarily depends on which apps or Dapps and services are busy and more connected with the public. And it doesn’t necessarily mean if a certain Dapp is on chain X, it could not also exist on chain Y. So I think it is a function of how successfully certain Dapp is in making its core product usable to the public. And with account abstraction, we expect Web3 Dapps to feel more like Web2. Hence, those that do have an intuitive user experience (UX) and user interface (UI) will stand to win the next phase of adoption.

Albert: I believe the developers mentioned earlier are deploying AA products on multiple chains, especially on layer 2. Based on our observations over the past few months, the majority of UserOperations have occurred on Arbitrum, OP, and Polygon. Well, even though Ethereum is the birthplace of ERC4337, the high gas fees on Ethereum have led usto believe that those layer 2 chains will be the first chains to witness widespread adoption of AA. But I need to add that we discovered an issue on Polygon. As mentioned earlier, bundler transactions sent on Polygon need to have MEV protection. Unlike layer 2 solutions with native MEV protection through sequencers, Polygon’s MEV ecosystem has been developing at a relatively slower pace. We’re looking for other solutions to protect the bundles’ transactions from front run. I don’t know if the send transaction conditional will eliminate this issue or not. We have not explored this in depth, but I think this is a great opportunity for us to learn this proposal and to see what will happen on Polygon in the bundler ecosystem.

Kristof Gazso: When it comes to the chains that are gaining traction, Pimlico’s focus is not solely on specific chains that we target. Our approach at Pimlico is centered around the idea that there are multiple decentralized ecosystems or worlds emerging, and our goal is to encourage these worlds to adopt smart account-based systems. By doing so, these worlds will naturally choose the chains on which they deploy their systems. We do not try to directly influence or favor specific chains. Instead, our approach is to support the success of these worlds, and if they thrive, they will likely deploy their systems on multiple chains.

Marc: I agree with Dror’s perspective that a significant portion of adoption will occur at the app layer. Currently, most application developers on the EVM primarily test their apps using the Metamask wallet and externally-owned accounts (EOA) without considering the possibility of integrating a contract wallet into their applications. In the near future, the networks or chains that see the highest adoption of account abstraction will be those that offer an enriched version of this functionality. Infinitism team, for instance, is actively working on a proposal to introduce ERC4337 as an enriched version for Layer 2 networks to adopt, while maintaining compatibility with the existing ERC4337 standard. By defaulting to or supporting the use of smart contract wallets, these networks can foster an ecosystem of applications that seamlessly interact with such wallets.

BlockPI Network is a decentralized infrastructure provider, offering one-stop services across 33+ networks and processing over 20B requests every month. Being a leading RPC provider, BlockPI’s RPC service is famous for its rock-solid stability, fastest response rate, and best price performance. By supporting ERC-4337 as a pioneer in this industry, BlockPI also is the only one providing the most diversified Bundler services in the market currently.

Website | Twitter | Telegram | Discord | Medium

About Moonshot Commons

Moonshot Commons is based in Hong Kong, Singapore, and New York, it is a global builder community for Gen-Z engineers in Web 3.

Website | Twitter | Medium

About Ethereum Foundation and ERC-4337 Community

Ethereum Foundation is a non-profit and part of a community of organizations and people working to fund protocol development, grow the ecosystem, and advocate for Ethereum. ERC-4337 Community is a community joined by numerous projects and developers who have made contributions to ERC-4337. ERC-4337 introduces Account Abstraction without any modifications to the core protocol. It also takes usability even one step further by introducing paymasters. Through ERC-4337 Account Abstraction is becoming widespread.

Eth Foundation Website | Eth Foundation Twitter | ERC4337 Community Website | ERC4337 Community Twitter

About StackUp

Stackup’s expertise in ERC-4337 is born out of practical experience and a relentless commitment to service excellence, evidenced by a >99.9% uptime. Choosing Stackup means opting for a collaborator who understands the complexities of account abstraction and provides robust, tailored services.

Website | Twitter | Discord

About Pimlico

Pimlico is an infrastructure platform for building the next generation of Smart Wallets. For developers who are developing an ERC-4337 Smart Wallet, Pimlico provide bundlers, verifying paymasters, ERC-20 paymasters, and much more to help them build a more stable wallet and iterate faster.

Website | Twitter

About Candide

Candide allows users to create a smart account and unleash the true power of Account Abstraction. Candide’s developer tooling leverages the power of every smart wallet seamlessly, eliminating the need for special integrations for individual wallets. With its unified approach, developers can effortlessly tap into the features of any ERC-4337 compatible smart wallet, streamlining development and ensuring a consistent experience for users across the entire ecosystem.

Website | Twitter | Discord

About Etherspot

Etherspot is an Account Abstraction SDK, delivering frictionless Web3 UX. Etherspot SDK represents a multi-chain self-custody Smart Contract Wallet platform that caters to the needs of decentralized application, game, and wallet developers. The platform offers seamless Web3 UX solutions that enable quick onboarding of new users while simplifying the intricate blockchain operations.

Website | Twitter | Telegram | Discord