In this article, I will be highlighting the Best ERC-4337 APIs for Smart Account Development in 2026. We will be taking a closer look at top infrastructures like Infura, Moralis, thirdweb, Biconomy, Alchemy, Safe, Stackup, ZeroDev and OpenZeppelin Defender, evaluating each one's set of features and benefits, limitations and shortcomings, underlying infrastructure and development capabilities in order to determine which one is best suited for your smart account development needs.

Welcome to our weekly digest, where we unpack the latest in account and chain abstraction and the broader infrastructure shaping Ethereum. **This week:** core devs hold off on making native account abstraction a Hegotá headliner after L2 pushback; a ZeroDev founder publishes a three-part retrospective on what ERC-4337 actually delivered; builders in the AA Mafia group debate why EOA migration has been slow; and the Ethereum Foundation drops its Poseidon hash in favor of standard hashes. * Native AA Decision Delayed as L2s Raise Concerns * ERC-4337 Retrospective: What Worked and What Didn’t * Inside the AA Mafia’s EOA Migration Debate * Ethereum Rethinks Poseidon for Its Post-Quantum Future Please fasten your belts! ## Native AA Decision Delayed as L2s Raise Concerns At All Core Developers Execution Call #243, developers finalized Glamsterdam’s gas repricing and moved its public testnet, Platåberget, forward, but the headline for us is what did not happen. A decision on making native account abstraction part of Hegotá was pushed back again. Client support for EIP-8141 Frame Transactions, the native AA enabler, has actually grown. Nethermind now backs it and could champion it as a headliner, Geth contributors call it increasingly important for post-quantum readiness, and Besu raised no opposition. The hesitation came from Layer 2s. Base and Arbitrum raised concerns about adopting Frames before the wider EVM ecosystem converges on one account model, warning it could fragment the stack if L1 and L2s head in different directions. There is already a competing design in play. Base has implemented parts of EIP-8130, an alternative that adds programmable-account features like key rotation and cross-chain account sync without new EVM opcodes, while Frame Transactions supporters argue their approach can address the concerns through further specification work. In its own Hegotá priorities paper, Ethlabs still ranks Frame Transactions as its preferred path to native account abstraction, and the proposal has advanced to “Considered for Inclusion.” Breakout calls resume on August 25, with a possible decision at the August 27 call. ## ERC-4337 Retrospective: What Worked and What Didn’t ZeroDev’s CTO, who led the development of the Kernel smart account, published a three-part retrospective following ERC-4337’s move to Final status. His blunt conclusion is that the standard succeeded, but not as the consumer-wallet revolution many expected. Production told a narrower story. Across roughly 1.2 billion UserOperations, sponsored transactions made up the vast majority of volume, activity concentrated on a few deployments like World Chain and Base, and there was little sign of people migrating their main EOA into a standalone smart account. In practice, ERC-4337 became a shared rail for gasless and embedded flows rather than a mass account switch. The Kernel history, from v1 to v4, sharpens the lesson. The capability that mattered most was not batching or sponsorship but replaceable validation, the idea that an account can outlive its key and swap in passkeys, recovery, or new signature schemes. Fine-grained permissions shipped as a “scalpel” but were mostly used as an on-switch, because the safe, narrow path was never the easiest one. His third piece argues the real unlock is in-place migration. The unit that should change is the authorization model, not the address, so users keep their existing account while its root credential is replaced, which is exactly what EIP-7702 starts and native account abstraction would finish. The hard part he flags is removing the old key completely. Even after delegation, the original ECDSA key can still authorize changes, and many apps still trust a raw signature through ecrecover, so a true transition has to fix both the protocol and the application layers. ## Inside the AA Mafia’s EOA Migration Debate A recent discussion among account abstraction builders in the AA Mafia Telegram group dug into a stubborn question: why has EIP-7702 adoption been slower than expected? The thread pulled in people from Base, the Ethereum Foundation, ZeroDev, Etherspot and Ambire, among others. The raw numbers are not small. One participant shared 7702 activation counts from roughly a week earlier, led by Trust Wallet at over 640,000, MetaMask at around 280,000, and Pimlico’s Simple7702 account near 200,000, with many more behind them. Still, most of those wallets do not enable 7702 by default, so a large base of plain EOAs remains. Builders disagreed on how to read this. A ZeroDev voice argued 7702 was designed for short-term EOA improvements and should be judged against the EOA transition rate rather than smart-account totals, while others worried Frame Transactions could hit the same slow adoption unless wallet teams commit to supporting them by default. A recurring theme was that the root key is both the strength and the weakness. As an Ambire participant noted, 7702 users cannot lose funds the way pure smart accounts can, because a root key always exists, but that same key is also the vector behind most drains and blocks a clean migration. Practical friction came up too, from delegated contracts not being available on every network to wallets restricting which chains can use 7702 and missing support for backend-triggered gasless flows. The takeaway echoes the retrospective above, that native AA needs the wallets and the packaging, not just the protocol. ## Ethereum Rethinks Poseidon for Its Post-Quantum Future The Ethereum Foundation is dropping Poseidon, the SNARK-friendly hash it had championed for years, from its plans for Ethereum’s base layer. According to researcher Justin Drake, future zero-knowledge designs will lean on standard hashes like SHA-2 or BLAKE instead. The reversal comes from a research breakthrough. New proof systems that work over binary fields can now make ordinary, battle-tested hashes almost as efficient inside a SNARK as the specialized ones, flipping the goal from SNARK-friendly hashes to hash-friendly SNARKs. The appeal is security through simplicity. Standard hashes have withstood decades of analysis, so relying on them avoids the risk of a newer, more exotic primitive being broken, and it removes the need to wait years for Poseidon to be battle-tested. Drake tied the move to a rough stretch for fancier cryptography, noting that lattice-based and isogeny-based schemes have taken hits in recent weeks, a thread we covered last week. For account abstraction, the same instinct favors conservative, swappable cryptography, which is one reason native AA through Frame Transactions is being designed to let accounts change signature schemes as the field shifts. **Start exploring Account Abstraction with Etherspot!** * Learn more about account abstraction here. * Head to our docs and read all about Etherspot Modular SDK. * Skandha — developer-friendly Typescript ERC4337 Bundler. * Arka — an open-source Paymaster Service for gasless & sponsored transactions. * Explore our TransactionKit, a React library for fast & simple Web3 development. * Follow us on X (Twitter) and join our Discord. ❓Is your dApp ready for Account Abstraction? Check it out here: https://eip1271.io/ **Follow us** Etherspot Website | X | Discord | Telegram | Github | Developer Portal

Learn about integrating smart accounts for Arbitrum and account abstraction.

Offchain Labs acquired ZeroDev, a company that manages over five million smart accounts and collaborates with projects like Infinex and DIMO.

The Bybit hack highlights a key design decision of EIP-7702 that is often criticized, but in fact protects users from this type of attacks. For context, the Bybit hack worked by tricking the multisig signers into signing a transaction that, while looking benign, in fact updates

World launched MiniKit 2.0 on March 31, aligning World Chain development with the EIP-1193 Ethereum standard so developers can ship once across web and World App, often with two lines of code. Transaction confirmation dropped from 2 seconds to 200ms via Flashblocks. Gas sponsorship via Privy and ZeroDev removes the wallet-funding barrier for new users. Seven local stablecoins were added. World Build 3 applications are open, with $20,000 in hackathon prizes and a build program that has produced $15 million in VC investment across previous cohorts.