2023 Recap: Modular Account Abstraction

2023 has been the year of the “account abstraction innovators”. ERC-4337 garnered the initial hype and kickstarted the developer community (the 4337 Mafia) but the impact has been far broader than the scope of the standard itself. New wallets have emerged (like Soul Wallet and Clave), vital smart contract account features have improved (session keys, new recovery mechanisms and @sevenxventures/webauthn-and-passkey-key-management-for-daily-crypto-users-2f61ced39eb0">passkeys) and adjacent verticals have found ways to leverage smart accounts to power other key UX improving infrastructure (e.g. intent-based transactions).

One emerging vertical that’s seen a ton of innovation, and sits right at the center of all these developments on top of 4337, is modular account abstraction. Modular account abstraction promises a world where smart accounts are easily extensible, customizable, portable and secure. We’re extremely excited about where this space is headed in 2024. To start off the year, we want to recap last year’s key milestones and briefly set out predictions for 2024.

Before 2023

The idea of modularizing smart accounts has been around for a while, with the most notable (and to our knowledge only) modular account prior to 2023 being the Safe. Its architecture allows for executor modules that call back into the account, a guard/hook module and a fallback handler. Some examples of modules built for the Safe include the Zodiac modules built by GnosisDAO. Other account abstraction players have also previously explored some aspects of modularity, such as Argent, but as far as we know have not yet transitioned to using this paradigm in production.

The starting gun

Late 2022, nearing the completion of the ERC-4337 audit, the idea arose of modularizing an ERC-4337 account using Diamond Proxies. At ETHDenver in February 2023, Rhinestone built the first prototype of such an account that was 1) modular, 2) natively ERC-4337 compatible and 3) based on Diamond Proxies and were chosen as one of the finalists in the hackathon. This sparked a lot of conversations, community efforts, prototypes and products.

Technical milestones / proposals

One of the first major milestones post ETHDenver was the launch of ERC-6900 (18 April), which proposed standardizing modular smart accounts and drawing inspiration from our early ETHDenver prototype. This formalized an open debate around the use of Diamond Proxies, and in particular delegatecall, to modularize smart accounts. ERC-6900 was soon followed by ZeroDev’s Kernel (25 April) which was and still is the most minimal, extensible smart account. Biconomy later launched their own modular smart account (27 Sep) by forking the Safe and making it natively 4337-compatible and allowing for modular validation logic. Thirdweb also announced the Dynamic Account in late summer.

However, modularizing the account is not the only piece of the puzzle. The promise of modular account abstraction is that any developer can build a smart account feature that can be installed and used by any smart account user. Introducing third party developers to the account creates large security risks. We launched the Module Registry (July), with ERC-7484 (14 August) as a quick follow, to address this specific problem and create a platform for module distribution and discovery. The Safe{Protocol} whitepaper was published soon after on 15 August to solve this exact problem within the Safe ecosystem.

Module development in 2023 primarily centered around account features that are table stakes. This included the development of novel account recovery systems (e.g. Safe’s Recovery Hub), improvements in the use of passkeys as a signing mechanism, and multi-chain modules to improve the synchronization of an account across multiple networks. Beyond these baseline account features, we began to see the greenshoots of more opinionated modules that provide users with greater security, privacy and easy access to DeFi directly from the account (more on this in predictions). To arm developers with the tools they need to easily use this modular architecture to build custom account features, Rhinestone launched ModuleKit (25 August), the first developer framework for module development.

In November, Biconomy and Rhinestone teamed up to announce the first “Module Store”, which is set to launch in Q1 2024. This Module Store allows dapp and wallet developers to discover modules built by others and integrate them into their application to improve the UX or create novel product experiences.

In December, ZeroDev, Biconomy, Rhinestone and OKX teamed up to launch a more minimal standard for modular smart accounts: ERC-7579. This ERC aims to bring about the minimal interoperability required by module developers and applications while also allowing account vendors to innovate, something that is severely restricted by ERC-6900.

Data insights

Modular accounts deployed

Despite the modular account abstraction ecosystem only gaining traction in 2023, the vast majority of accounts in production are already modular. The Kernel and Biconomy account reflects over 80% of the 4337-native accounts and Safe is the market leader in terms of smart account adoption. The Kernel has deployed almost 1m accounts whilst Biconomy has deployed over 600k (reference). In 2023, Safe continued their dominance, deploying over 4.1m accounts, a 355% increase YoY. Many of these accounts have been deployed on behalf of Worldcoin who earlier this year reached 2.7m verified accounts.

Apps utilizing modular smart accounts

Within the 4337 ecosystem, the largest applications by number of accounts and userOps are CyberConnect and FanTV (a decentralized media startup), with 550k and 470k unique accounts respectively. The ZTX hoodie drop generated over 200k accounts and Grindery, a telegram smart wallet, has created over 200k accounts. All these applications are utilizing token incentives to convert users, making it difficult to comment on early product-market-fit (the impact this has had on user engagement can be seen in the next section). These products have also primarily used smart accounts to create a seamless onboarding experience, which is not unique to modular account abstraction.

An area where modular smart accounts provide a more novel application experience is DeFi via account automation which is emerging as the platform of choice for the intent-based transaction infra and product builders. For example, Composible CoW employs Safe’s modular architecture to create intent-like transactions for users. Rage Trade and CapX provide seamless DeFi actions through purpose-built session keys. However, these products are still early experiments and have yet to fully tap into smart account modules.

User stats

One major shortcoming of the ERC-4337 ecosystem as a whole so far has been very low user retention rates, as seen on the chart below (from BundleBear). Because the vast majority of ERC-4337 accounts are modular, the same is true for the modular account abstraction ecosystem. As mentioned above, a large driver of this is likely the result of the token incentives that have been utilized to drive initial user adoption of the affiliated product.

Source: bundlebear.xyz

Predictions for 2024

We believe the two keywords for 2024 will be adoption and interoperability.

Adoption

In 2023, we’ve seen most smart account adoption happening through embedded wallets or, what some have proposed to call them instead, embedded signers. The vast majority of smart accounts that are controlled by these signers are modular and we think that this trend will only continue in 2024. The rise of new embedded signer SDKs and improvement of existing ones will push builders to keep competing to offer new features to their developers, and we believe smart account modules will play a key role in creating differentiation between players.

Adoption of existing modules, such as passkey verifiers or various recovery and session key types will continue to grow. Developers will also experiment with more unique modules, such as DeFi automation executors, that will help them deliver unique product experiences to rival or surpass those of traditional and centralized finance. Modules could play a key role in bringing bots onchain for any given user and drastically reducing the knowledge barrier for DeFi adoption.

We also see module adoption within the areas of security and privacy. Spending limits and configurable multisig (e.g. requiring an extra signer if interacting with a new contract) provide simple security enhancements, but more advanced exploit detection modules can be coupled with offchain services. Modules that utilize privacy technology will emerge to provide users with a means to obfuscate their identity and increase privacy. This could be as simple as ensuring the account recovery process does not dox the user or providing novel mechanisms for stealth P2P interactions.

Interoperability

We’ve seen a lot of fragmentation in 2023 as different players have paved their own path to innovate around modular account abstraction. While we think the innovation will continue, we also think that interoperability will play a bigger role in 2024, allowing applications and users to more easily use modules and developers to distribute their modules far more easily to any user regardless of what account implementation they use.

We see three areas in which interoperability is likely to happen: 1) on the account contract level, 2) on the account creation level and 3) on the module level. ERC-7579 is the best candidate for the first of these, with ZeroDev, Biconomy and OKX (together currently more than 80% of all 4337 compatible smart accounts) adopting it in Q1 and more account builders following. The second will aim to make it easier to both create and re-use embedded signers across multiple applications powered by the same SDK and, perhaps, also across multiple SDKs. In 2023, we’ve already seen one example of a standard that aims to achieve this, ERC-7555, and multiple talks at WalletUncon during DevConnect were focused on combating this problem. Finally, we think that 2024 will be a big year for module interoperability, both on a high level such as proposed by ERC-7579, and on a more module-specific level, such as standardizing how, for example, session keys work.