加密錢包革命：帳戶抽象如何改變一切

簡單來說，帳戶抽象（Account Abstraction）讓用戶可以將智能合約作為自己的帳戶——真正令加密錢包變得可編程。這是用戶與區塊鏈應用互動方式的重大轉變，亦被認為是令加密貨幣向用戶友善、安全及大眾採用的關鍵步伐。

Ethereum’s 聯合創辦人Vitalik Buterin曾表示，若不採納帳戶抽象，Ethereum或未能實現其目標，由此可見這技術對Web3未來的重要性。

但究竟甚麼是帳戶抽象？又如何運作？要了解其重要性，我們要先明白區塊鏈傳統的帳戶運作方式，以及這種模式有何限制。然後，我們會深入探討帳戶抽象如何改寫規則，發揮哪些優勢（如提升保安與易用性）、現實例子，及餘下挑戰。

最後你會明白為何帳戶抽象被譽為對加密錢包的重大升級——有望令管理加密貨幣變得像用現代金融App一樣簡單流暢。

傳統帳戶模式：EOA對比智能合約帳戶

如今，以太坊等區塊鏈採用帳戶模型來管理資產及執行交易。以太坊上的帳戶有兩大類：

外部擁有帳戶（EOA） —— 即「普通」用戶帳戶，由個人私鑰控制。你若開設過以太坊錢包（如MetaMask或硬件錢包），便擁有一個EOA。EOA有公開地址（由公開密鑰衍生）及私鑰用來簽署交易。持EOA可存幣、傳送資金或調用智能合約功能。EOA可自行發起交易（靠私鑰簽名），但不能執行自訂代碼——即功能僅限基本轉帳。EOA主要有兩個功能：發送資產（ETH或代幣）至其他帳戶，或呼叫智能合約內的某個功能。
合約帳戶（智能合約） —— 此類帳戶由代碼（智能合約程式碼）支配，並非由私鑰控制。合約帳戶可持有資產，也能自訂規則或邏輯（程式碼），在收到交易時自動執行。比如分散式應用或代幣合約都存在於合約帳戶。然而，合約帳戶不能自行發起交易，必須由EOA或其他合約觸發才能執行其代碼。即是說，必須有人（或外部帳戶）叫用該合約才有動作。合約帳戶可完全自訂，支援複雜邏輯，但無法自主發起交易——沒私鑰可直接控制向外送出交易。
以太坊帳戶類型對比：外部擁有帳戶（EOA）可用私鑰控制並發起交易，但不能執行自訂程式碼；智能合約帳戶（SCA）可執行程式碼，但不能自主發起新交易。帳戶抽象旨在消除這種分隔。*

目前這兩類帳戶相互分開，各自有不足。EOA綁定單一私鑰，局限大且高風險：失去該私鑰（或備份助記詞）即失去帳戶及資產——區塊鏈沒「忘記密碼」可用。被惡意人士取走私鑰，對方可全數盜走資產。EOA無內建消費上限、多重簽名機制、多方協助復原存取等傳統銀行防詐保障。EOA每次交易都需手動簽名，也需帳戶有足夠原生幣（ETH）支付汽油費，對普通用戶既不便又高風險。正如Rumble Fish開發團隊所說：「EOA沒社交復原、沒消費限制、沒2FA、又必須設有ETH支付費用——新手體驗糟糕。」

智能合約帳戶則靠程式碼帶來彈性。 例如，可設計成多人才可批准交易（多簽錢包）、可設每日提現上限，抑或支援「社交復原」機制（讓信任朋友協助恢復存取權）。這些在EOA無法實現，於是有例如Gnosis Safe（多簽錢包）、Argent（支援社交復原錢包），它們通過合約帳戶加入額外安全功能。不過，合約帳戶始終不能自主發起交易，通常還是需靠EOA帶動。以智能合約錢包為例，常須有中介服務（外部EOA）協助將用戶指令包裝成EOA交易，再送至區塊鏈。你需給這些中介繳費，或要因為支付汽油費而又持有額外ETH。設計失當的合約型錢包，操作甚至比普通錢包更繁複。過去要用智能合約帳戶，往往還須額外備有EOA賬戶或服務支援，令用戶體驗增加阻力。

總結來說，現行模式下：

EOA = 你的錢包（由單一私鑰控制）——簡單但不彈性且無容錯。
合約帳戶 = 可編程金庫（進階功能）——強大但非自動。

帳戶抽象希望打破這障礙。願景是讓用戶帳戶像智能合約般靈活，而無需依賴額外EOA才能運作。即是讓每個帳戶都「變聰明」。如此，用戶可隨心自訂帳戶安全與操作，而毋須捨棄發起交易的能力。讓我們來看看帳戶抽象如何做到這點，為區塊鏈用戶帶來徹底體驗提升。

甚麼是帳戶抽象？

核心是：帳戶抽象旨在統合上述兩種帳戶，令用戶帳戶行為如智能合約。即由傳統只遵從協議內建規則的EOA，變為用戶可將消費及安全邏輯寫入並完全控制的智能合約。實際上，錢包可成為一個於區塊鏈上的智能合約，專門針對該帳戶定制認可什麼為有效交易的規則。用戶可透過直接或以錢包友善功能自訂這些規則。

另一個定義：帳戶抽象通過「抽象化」標準交易驗證規則，讓自訂驗證邏輯有彈性，令用戶帳戶更靈活。目前，以太坊每宗交易必須滿足一些內置條件，包括由正確密鑰（ECDSA）簽名、交易序號正確，及有足夠ETH支付汽油費。全體EOA皆然。帳戶抽象則希望鬆綁這種一刀切規定。即是每個帳戶不再強制只用同一組簽名方式（單一密鑰ECDSA）、或必定付ETH手續費，而是可逐個帳戶自訂有效性條件。基本上，交易驗證程序本身可以寫成智能合約：只要你的交易（或指令）符合同帳戶合約規定的條件，即屬有效。

這帶來無限可能。例如，應用帳戶抽象可以打造：

多擁有者帳戶——需多方簽名（多簽）或其他條件共同批准交易
後量子安全帳戶——可用新型抗量子運算簽名方法（非僅限ECDSA）
毋須任何簽名的帳戶（如某些特殊應用場景），或利用可信模組進行生物認證等
可用不同資產支付汽油費、甚至由第三方代付汽油費的帳戶——毋須帳戶長期存ETH
公開或限時帳戶——需在某時段後、或任何人達到指定條件可執行（想像可被任何人執行的遺囑）

總括而言，帳戶抽象解放用戶自訂錢包規則，不再受制於區塊鏈的底層規矩。這種模式將「用戶帳戶」及「智能合約」二合為一。以太坊研究員Ansgar Dietrichs就此稱：「讓『可編程錢包』成真」——你的錢包本身就可包含安全性、恢復、批量處理等邏輯，而毋須外部軟件才能做到這些功能。

為何這件事如此重要？ important?** 因為許多阻礙加密貨幣大規模採用的痛點，都是起於EOA（Externally Owned Account，外部擁有帳戶）的局限。新手用戶經常難以安全地管理私鑰和助記詞，萬一發生錯誤也完全沒有補救措施。有經驗的用戶則擔心單點故障問題 — 一旦私鑰遭駭，後果極其嚴重。開發者一直被迫構建一些笨重的權宜之計（例如 relayer 網絡或中心化服務），才可以提供如免GAS交易、社交恢復等功能，因為區塊鏈本身並未原生支援這些功能。Account Abstraction（帳戶抽象）正面解決這些問題，讓帳戶模型本身變得更強大和以用戶為中心。因此，它被視為Web3下一階段進化的關鍵基礎設施。事實上，帳戶抽象多年來一直是以太坊開發者，包括Vitalik Buterin等人夢寐以求的目標，他們在多個場合倡議帳戶抽象，作為極大提升以太坊易用性和安全性的路徑。如今，這不再只是紙上談兵 —— 透過新標準，帳戶抽象正於以太坊變成現實，而更新的區塊鏈則一開始就用上帳戶抽象設計。

為了更好理解由理論到實踐的過程，不如一起看看以太坊具體如何實踐帳戶抽象 —— 特別是透過一項稱為ERC-4337的升級 —— 並了解其底層運作原理。

Account Abstraction於以太坊（ERC-4337）的運作方式

以太坊推進帳戶抽象的過程，最近達到了一個里程碑，就是ERC-4337（亦稱為EIP-4337）的提案。這提案於2021年提出、2023年上線，為以太坊帶來了帳戶抽象，而無需要對以太坊核心協議做出根本性更改。這點很重要，因為L1核心協議的修改程序緩慢，且需要極大共識。相反，4337善用智能合約及鏈下基礎建設於以太坊之上實現帳戶抽象 —— 這個巧妙方案讓我們可以馬上享受到AA（Account Abstraction）的好處，毋須硬分叉。

究竟它如何運作？ERC-4337 定義了一個嶄新的交易流程，重心在於"User Operation"物件（簡稱UserOp）。User Operation像是一個包裝好的交易，代表用戶的智能錢包所想執行的操作。與其直接由用戶錢包發起一個普通以太坊交易（如EOA執行的），智能錢包會生成一個包含所有動作細節的User Operation：包括發起人（sender）是誰、操作目標（可能是合同調用或代幣轉帳）、調用的數據和參數、以及相應的驗證簽名或憑證。

以下是ERC-4337的概覽流程及相關新組件：

User Operations & Mempool（用戶操作與緩衝池）： 當使用支援ERC-4337的智能錢包時，錢包並不會廣播一個普通交易，而是組成一個UserOperation物件及簽名（而且這個簽名可以是任何合約邏輯預期的東西，毋須只限於一把EOA密鑰）。這些UserOps會進入一個專用的UserOperation Mempool，與普通以太坊交易的mempool分開。你可以把它想象成一個"預備區"，讓各種智能合約錢包的操作等候被打包。
Bundlers（打包者）： 接住來的是bundler（打包者），他們的角色有點像「礦工」或「區塊產生者」，但只對應用戶操作層級。一個bundler節點會監察這個UserOp mempool，然後集合多個用戶的UserOp打包成"bundle"，再包裹成一個以太坊L1的交易。簡單來說，bundler代表多個用戶共同向區塊鏈提交操作。Bundler本身必須是EOA（因為現時協議下只有EOA可發起L1交易）—— 但終端用戶無需每人都動用EOA交易。Bundler會為這宗大交易先墊支Gas費，然後向每個UserOp收取手續費作補償。
EntryPoint Contract（入口點合約）： 被打包後的交易會發送到以太坊一個特定的EntryPoint智能合約。這就是ERC-4337設計的核心組件。EntryPoint合約會驗證並執行bundle內的User Operations：解包該bundle後，對每個UserOp, 它調用目的地的智能合約錢包（即用戶的帳戶合約），作相應操作的驗證與執行。每個智能合約錢包必須實現EntryPoint調用的標準介面，一般包括如 validateUserOp（驗證簽名、nonce及該帳戶規則）及 execute（驗證通過則執行所需動作）等函數。如果有UserOp驗證不通過（例如簽名錯誤或資金不足），EntryPoint會拒絕執行該非法操作。
Paymasters (可選) 支付主模式： ERC-4337同時引入Paymaster概念，即輔助智能合約，可資助Gas費或指定Gas由誰支付、如何支付。如果有用到Paymaster，用戶的UserOp可以關聯一個Paymaster，在EntryPoint驗證時，由Paymaster為用戶支付Gas（有時會根據某些條件確認後才執行）。這機制讓用戶無需持有ETH也能進行交易 — 例如dApp開發者可以設Paymaster幫新用戶支付手續費作友善引導，或可直接以用戶持有的ERC-20代幣支付Gas。如果沒有Paymaster，Gas就從用戶自己的智能錢包資金扣除（如果錢包有內部兌換或相關邏輯，甚至形式也可以是ERC-20代幣）。
Bundler Incentive 打包者獎勵： 運行完所有操作後，EntryPoint合約會用用戶帳戶或Paymaster提供的資金支付bundler應得的手續費。這激勵bundler繼續運作。本質上，bundler賺取手續費的方式與礦工/驗證人收Gas費類似，但如今是批量處理多個用戶操作，一次賺取。

這個架構把每個用戶必須擁有EOA的需求徹底抽象走了。只有bundler需要用EOA提交交易，其餘所有人的“交易”都是UserOp，被合約管理。如Rumble Fish團隊形容，在4337模型下，“bundler是這個帳戶抽象生態圈唯一需要[EOA]的參與者”。而最終用戶的帳戶變成純粹的智能合約錢包 —— 用戶從來不用親自發送EOA交易，他們的指令依然可透過EntryPoint在鏈上執行。

不如舉個簡單例子幫助理解：假設Alice有一個支援帳戶抽象的智能合約錢包，設定了規則“每日最多只允許我朋友Bob由我的錢包花出0.1 ETH”。這是普通EOA做不到的 —— EOA本身無法在鏈上設有限度委託。現在, Bob希望在Alice離線時幫她執行一個交易。Bob發起一個UserOperation，調用Alice的合約錢包：“由Alice轉出0.05 ETH到某個DEX”。Bob簽署這個UserOp（可能用他自己的key，只要Alice合約寫好授權就可以）。UserOp進入mempool，bundler收集後與其他操作一併送去EntryPoint，EntryPoint調用Alice錢包合約的驗證函數，合約會檢查“Bob是否有權限？這金額是否在0.1 ETH每日限額之內？”如果通過，驗證成功。然後EntryPoint叫Alice錢包的execute函數，把0.05 ETH轉給那個DEX，操作順利完成；EntryPoint隨後會用Alice錢包或Bob或Paymaster的資金支付bundler一些gas費。期間，Alice完全不用操作 —— 她事先設好錢包規則就行。如果Bob嘗試超額或沒授權，合約在驗證時自然會拒絕操作。

由此可見，帳戶抽象確實帶來不少新組件（UserOp、bundler、EntryPoint、paymaster等），但對用戶來說，這些全都可以是底層實現細節。用戶只會感受到更靈活的錢包體驗。正如有開發者所述：“最終用戶根本不會意識自己在用Account Abstraction，他們只見到以前無法實現的功能。”當然，這要安全落地並不簡單，例如EntryPoint合約就成為十分關鍵的基礎設施，必須零漏洞和受信任。（幸好它已經過審計及形式驗證。）每個智能錢包代碼也要妥善管理簽名、nonce及安全檢查。所以，要讓AA安全無縫，背後需要大量工程心血。

值得留意的是： 以太坊並非唯一追求帳戶抽象的網絡。一些較新的區塊鏈和Layer-2已經較容易原生整合帳戶抽象。例如StarkNet（一個以太坊的Layer-2 zk-rollup），設計時就預設每個帳戶本身就是智能合約帳戶，完全沒有EOA的概念 —— 連用戶帳戶都有代碼（一般是標準帳戶合約，可升級或配置以滿足用戶需要）。這即代表StarkNet一開始就有了AA功能（如社交恢復等），全然不用像4337般靠外部bundler，sequencer本身已原生處理智能帳戶。類似地，zkSync v2（另一個Layer-2）也在協議層面原生支援帳戶抽象 —— 為每個用戶部署一份默認帳戶智能合約，模仿EOA行為（除非自定義），從而保持兼容但仍可擴展。這些路線都驗證了power of AA: by removing the EOA/contract split at the protocol level, one can design much richer wallet functionality. Ethereum’s ERC-4337 approach is a bit more elaborate (since it’s layered on), but it ultimately provides the same end result on Layer-1 Ethereum.

AA嘅威力：由於喺協議層面移除咗 EOA 同智能合約 wallet 嘅分隔，可以設計更加豐富、進階嘅錢包功能。雖然 Ethereum 嘅 ERC-4337 實現依賴疊加方式，結構上複雜啲，但最終都係喺 Layer-1 Ethereum 上帶嚟同一種效果。

Now that we have a grasp of what account abstraction is and how it functions (at least in Ethereum’s implementation), let’s turn to the benefits. Why all the buzz? What does this actually enable for users and developers that wasn’t possible (or was very difficult) before? The advantages of account abstraction are numerous, touching security, usability, and beyond.

而家我哋大致明白咩係 Account Abstraction（AA）同埋依家操作方式（最少喺 Ethereum 上嘅例子），我哋可以進入重點：到底有咩著數？點解咁多人討論？究竟 AA 真正為用戶同開發者帶嚟乜以前冇可能（或者極難實現）嘅功能？AA 嘅優點好多，涵蓋安全性、易用性、同埋更多層面。

Benefits of Account Abstraction

Account Abstraction 優勢

Account abstraction is often touted as a game-changer for user experience and security in crypto. By letting wallets become smart contracts, it unlocks features that make managing crypto more akin to managing a modern bank account or online profile, rather than handling raw cryptographic keys. Let’s break down the key benefits:

AA 經常被視為改革加密貨幣安全同體驗嘅重要一步。令錢包變做智能合約，開放更多功能，令管理加密貨幣好似管理現代銀行帳戶或者線上 profile 咁，而唔係要用家親手處理私鑰。以下分拆主要優點：

Improved Security and Recovery Options

更高安全性及恢復方式

One of the strongest appeals of account abstraction is the potential for dramatically improved security** of crypto accounts. Today, if you lose your seed phrase or private key to an EOA wallet, you simply lose access – there’s no recovery. Likewise, if your key is stolen, the thief can drain everything, and there’s no one to call to freeze the account or undo the damage. This harsh reality has led to countless tales of lost fortunes and remains a top fear for new users.

AA 最吸引之處之一，就係大大提升加密錢包嘅安全性。依家如果你唔見咗自己 EOA 錢包嘅 seed phrase 或私鑰，就會永久鎖死自己個錢包，冇辦法恢復。又或者私鑰俾人偷咗，對方可以即時清袋，冇人可以幫你凍結戶口或者追得番。呢種殘酷現實令無數人損失巨款，亦係最多新用戶最擔心嘅事。

Account abstraction offers a fix: because your account is a programmable contract, you can build in your own security mechanisms. For example, developers can implement smart wallets with social recovery or multisig approval. In a social recovery wallet, you still have a primary signing key for everyday use, but if you lose it, a group of “guardians” (friends, family, or even other devices of yours) can collectively authorize a replacement key for your wallet. This means no single point of failure – losing one key doesn’t lock you out forever, and no single stolen key (unless all guardians are compromised) lets an attacker in. Vitalik Buterin has championed social recovery as his preferred way of securing wallets, and with account abstraction this model becomes much easier to deploy widely (indeed, projects like Argent have used a form of it via smart contracts).

AA 其實帶嚟咗解決辦法：由於你個戶口其實係個智能合約，可以自訂唔同安全機制。例如，開發者可以整啲「社交恢復」或者多重簽署（multisig）認證嘅智能錢包。如果用咗社交恢復（Social Recovery），你日常都會用返自己的 key，但萬一唔見咗，可以叫一班守護者（可能係朋友、家人，甚至你自己其他裝置）幫你啟動更換新 key。變相唔會有單一失誤就令戶口全軍覆沒，亦唔會單靠一條被盜私鑰就令黑客得手（除非全部守護者都出事）。Vitalik Buterin（以太坊創辦人）一直都推薦用 social recovery 做錢包安全，AA 令呢套方案更加容易普及（好似 Argent 咁已經有智能合約錢包用呢個模式）。

Similarly, account abstraction allows multisignature wallets to become mainstream for individuals, not just organizations. You could require that any transaction from your wallet needs to be signed by, say, your phone and your laptop (two keys) – which greatly reduces the risk of a single device compromise. In the past, multisig wallets existed (like Gnosis Safe) but were mostly used by teams or experts due to complex setup. With AA wallets, a user-friendly interface can let anyone toggle on a 2-of-3 multisig for themselves, or add a daily spending cap that requires an additional confirmation if the limit is exceeded. These kinds of custom rules were not possible with plain EOAs.

同樣，AA 令多簽 wallet（multisig）的應用，由以往只有組織先用，開始普及到一般個人用戶。你可以設定，所有交易都要你手機同電腦一齊簽名（即兩條 key），先可以完成，大大減少個別裝置失竊所帶來嘅風險。以前 Gnosis Safe 咁嘅 multisig 主要係團隊同專業用家使用，設置又複雜；AA wallet 可以用一個 user-friendly 界面，俾任何普通用戶一鍵開啟 2/3 多簽，或者可以加埋每日消費限額（超額就要多一次確認）。好似呢啲自訂規則，以前 EOA 做唔到。

Crucially, account abstraction means developers have free rein to “get crafty and program all sorts of options for account authentication and recovery,” as one analysis noted. Want to add a two-factor authentication (2FA) where a mobile device must co-sign transactions? You can. Want the wallet to have a “freeze” function so you can lock it if you suspect a hack (sort of like freezing a credit card)? You can code that in, too. How about whitelisting certain “safe” addresses (like your own cold storage) that can receive unlimited funds from your wallet, but require extra checks to send to others? All doable with contract logic. In short, account abstraction brings the flexibility of modern security design to crypto wallets, which until now were stuck with an all-or-nothing key model. This greatly reduces many of the vulnerabilities and points of failure that plague EOA wallets. Users no longer have to walk a tightrope without a safety net – lose one key, you might still have other ways to recover; see a suspicious attempt, you might have programmed circuit-breakers.

最重要嘅係，AA 俾開發者有彈性發揮，可以「花心思寫各式帳戶認證同恢復方案」。例如想加多步二次驗證（2FA），要手機裝置一齊 co-sign 交易？冇問題。想錢包有「凍結」功能，懷疑被入侵時可以 lock 住成個戶口（好似信用卡咁凍結）？照樣可以寫 code 做到。甚至可以設自動白名單，指定某啲「安全」地址（例如你冷錢包）可以無限收款，向其他人出款就要更多檢查？智能合約都可以照程式邏輯做。總結都係：AA 為錢包引入現代安全設計嘅彈性，不再局限於「有 key 就用到，冇就完」咁死板。咁啲傳統 EOA 嘅各種弱點、失敗風險，大大減少。用戶唔使再踩鋼線，一失手即萬劫不復，可以有備份、有後著、有安全閘。

Lower Barrier to Entry for New Users

降低新手入門門檻

Beyond security, account abstraction can make using crypto much more accessible to everyday users. Let’s face it: managing an EOA with gas fees and seed phrases is intimidating for newcomers. The UI/UX has been often compared to the early days of the internet – you effectively ask users to manage secret keys (like long passwords) perfectly, and to grasp concepts like gas and nonces from day one. This is a barrier to adoption.

除咗安全，AA 亦令日常用家更容易入手 crypto。不如講真，EOA 加 gas費、seed phrase，普通新手一見到已經頭痛。UI/UX 好似返咗互聯網出世早期：新用戶要學識管理複雜長密碼（其實就係私鑰），又要即刻明白咩叫 gas、nonce 等專有名詞。呢啲全都係阻住新用戶入場嘅門檻。

Account abstraction lowers this barrier by enabling more familiar, user-friendly experiences. For instance, with paymasters covering gas fees or allowing gas in stablecoins, a new user could execute their first blockchain transaction without even owning ETH for gas. A dApp or wallet could sponsor the gas fee (perhaps as an onboarding promotion or using a freemium model) – the user just sees their action go through, akin to how a fintech app might waive fees for a first transaction. This is huge: requiring new users to first acquire ETH (often on an exchange) just to be able to use a dApp has been an onboarding nightmare. Account abstraction removes that hurdle by allowing gas fee abstraction – users can pay in whatever token they have, or not pay at all if a third party steps in.

AA 可以大大降低門檻，因為令用戶體驗變得似返平時 App。好似有啲「paymaster」可以幫用戶代付 gas費，甚至可以用穩定幣做 gas。即係新手連 ETH 都未必需要有，可以行第一次交易。dApp 或 wallet 可以做推廣，搵第三方 sponsor gas（好似一般金融 app 規限首單免費咁），用戶感覺係一 click 就做到操作。呢個改變好重要，因為以前人要去交易所搞 ETH，只係為咗用下某個 dApp，經常搞到 onboarding 完全失敗。AA 令 gas 抽象化，用戶可以用手頭任何 token 俾 gas，甚至唔使自己俾錢。

Another user-experience improvement is the idea of “signless” or one-click transactions. While not literally without a signature (there is still cryptography under the hood), from a user perspective, you could be “logged in” to a dApp for a session and not have to manually confirm every action. With account abstraction, wallets can implement session keys – ephemeral keys that have constrained rights (e.g. they can perform only certain actions for a limited time). You could log in to a game dApp and approve a session key that allows that game to execute moves on your behalf for, say, the next hour, with a cap on spending. During that hour, you enjoy a seamless experience as if you were just playing a normal online game – no transaction pop-ups every minute. Your wallet’s smart contract ensures the session key can’t do anything beyond its given permissions, and after an hour it’s void. This kind of flow is analogous to how web2 apps maintain sessions, and it’s enabled by account abstraction’s flexibility. Early implementations of session keys and “login with Ethereum” experiences are being explored now using AA wallets.

另一個 user experience 大提升就係「無需確認」或者「一鍵」交易。雖然背後都仲要 cryptographic 簽名，但表面上你可以「登入」一個 dApp session，之後嘅所有行為都唔使逐次手動 confirm。AA wallet 可以支援 session key，即用一條臨時 key 幫你實行有限時間/功能嘅操作。例如你登入 game dApp，批咗一條 session key，呢個 session 內可以俾個遊戲自動幫你操控（限住時間/金額），一個鐘後自動失效。咁用戶玩遊戲時就好似普通手游/網頁遊戲，唔駛每分鐘彈個 metaMask 確認。你個智能合約 wallet 保證 session key 唔會有多餘權限，時間到就失效。咁嘅 flow，就同 web2 session 好似，正正係 AA 嘅彈性先做到。依家 AA wallet 開始測試 session keys、Ethereum 登入咁功能。

Furthermore, account abstraction can allow features like automatic payments or subscriptions. As mentioned earlier, Visa’s crypto research team demonstrated a proof-of-concept where a smart contract wallet could execute recurring payments (auto-debits) on its own schedule. In their scenario, a user could schedule monthly bill payments from a self-custodial wallet – something that currently only custodial services or centralized banks can do – by giving a smart contract pre-approval to pull funds when due. This was done on a Layer-2 (StarkNet) with native AA, but the concept applies broadly. Imagine being able to schedule trades, bill payments, or transfers in advance with conditions (“execute this transaction if my balance is above X on date Y”) – it becomes possible when your account can execute code autonomously when triggered appropriately. The user doesn’t need to be online clicking “confirm” every time; their wallet contract acts according to rules they set.

仲有，AA 令自動付款（自動轉賬、訂閱扣費）變到做得到。好似早前講過，Visa 加密團隊 demo 咗智能合約錢包自動定期付款。即用戶可設定月繳水電費、定時自動 debits，用自託管錢包都做到 —— 呢啲以前一定要經 custodian/custodial services 或傳統銀行。當然，早期例子係喺 Layer-2 （StarkNet 原生 AA）示範，個 concept 可以全面應用。想像吓，可以預先訂定「到某日戶口夠額就自動付款/轉賬/交易」（即指令組合，自動觸發），用戶唔需要次次 online confirm，自動按規則執行。

All of these improvements lead to a friendlier onboarding and usage experience. One blog aptly noted that with account abstraction, dApps can start to feel as smooth as traditional fintech apps, where users don’t worry about the arcane blockchain bits like nonces and gas. Users could recover accounts via familiar processes (contact guardians or use backup devices, analogous to “reset password”), and they can use apps without understanding gas (the complexity is handled under the hood, possibly by the app). For non-crypto-native folks, this is a big deal – it’s the difference between requiring someone to configure their internet via command-line vs. just tapping an app icon and using a service.

總括而言，上述各項改良，令冇玩過 crypto 嘅新用戶 onboarding 體驗好好多。有人形容，用咗 AA 之後，dApp 登記/登入/操作，好似傳統金融 app 咁順手，唔駛再去理咩 nonce、gas。用戶可以用熟悉嘅流程重設、恢復帳戶（即搵開 guardian 或 backup 設備，類似 reset password），完全唔需要明或者理 gas（App 幫你包底搞掂晒）。對新手嚟講，呢個差距，好比要你用 command line 設定 wifi，同掂個 App icon 用服務咁大分別。

Customization and Automation of Transactions

交易自訂化及自動化

With account abstraction, users gain more control over what their accounts can do – and can automate complex tasks that previously required manual effort or trust in external services. We touched on some examples, but let’s highlight a few key capabilities:

AA 俾用戶對自己戶口可以做啲乜有更大主導權，仲可以自動執行啲以前要靠人手或者第三方服務先做到嘅複雜操作。上文都有提及，不過而家集中講一啲重點功能：

Batching and Complex Actions: Traditional EOAs must submit one transaction at a time, each with a separate confirmation. Smart contract wallets can be designed to batch multiple actions into one meta-transaction. For instance, you could atomically execute a series of steps: trade on a DEX, then lend the proceeds on a lending platform, then transfer the obtained token – all in one transaction from your smart wallet. This not only saves time and clicks, but can save gas by combining steps. Indeed, one cited benefit of account abstraction is the ability to “batch multiple transactions together,” reducing overhead and possibly saving fees. For users, it means one-click strategies instead of juggling multiple tx.
**交易批量處理及複雜組合操作：**傳統 EOA 每次只可以下一個交易、逐個 confirm。用智能合約 wallet 可以設計「批量」操作：多個 action 一齊喺一個 meta-transaction 入面完成。例如可以：喺 DEX 買幣 → 經 lending platform 借出 → 再轉出新 token，全部一單交易做晒。咁唔單只慳時間、減少 click 次數，仲可以結合步驟節省 gas。本身 AA 嘅好處之一就係可以「batch」執行多單交易，減攤 overhead 或 potential 手續費，令用戶可以一 click 完成複雜組合操作，唔駛自己分開做。
Pre-Approved Transactions & Automation: You can authorize certain transactions to happen under certain conditions without further approvals. This can enable things like stop-loss orders in DeFi (your wallet auto-executes a trade if price hits a threshold), or gameplay moves in blockchain games that auto-execute within certain parameters. Because the account is running your will as code, it’s like having a personal agent on-chain. A concrete real-world use: someone can program “if I don’t interact with my account for 1 year, automatically transfer funds to my backup wallet” – a kind of deadman switch, providing an inheritance mechanism. Without AA, this requires either trusting a third-party
**預先批核及自動化執行：**你可以設定某啲條件之下自動完成交易，唔駛每次人手批核。例如：喺 DeFi 做止蝕單（當價格到咗預設位錢包自動交易）、又或者區塊鏈遊戲內，只要在遊戲參數守則之下，會自動行動。其實即係你個戶口化成寫咗 code 嘅「個人代理」，乜 script 都可以寫。一個實例係：你可以寫低「如果我一年冇登入錢包，啲錢自動轉去 backup 銀包」，即係類似 deadman switch，做遺產/保險安排。冇 AA 的話，就要相信第三方託管先得。service or hoping someone calls a special contract; with AA, your own account can enforce it.

服務，或者希望有人去呼叫一個特定合約；但用 AA，你可以用自己嘅賬戶直接執行。

Easy Asset Management: Account abstraction allows features like “transfer all my tokens with one function call”. Normally, if you wanted to migrate to a new wallet, you’d have to send each token one by one. A smart wallet could have a method to sweep all assets (ETH and any tokens, NFTs, etc.) in one go to another address – simplifying migrating wallets or consolidating assets. It could also allow changing ownership of the wallet itself: for example, selling a wallet or transferring it to someone (something not straightforward with EOAs, since an EOA is tied to a fixed key that you shouldn’t share).
輕鬆資產管理： Account abstraction（賬戶抽象化）可以實現「一個動作就轉走晒所有代幣」咁嘅功能。以往如果你想換去新錢包，通常要一粒一粒token逐個轉走。而一個智能錢包就可以有一個方法，一次過「掃走」所有資產（包括ETH、任何代幣、NFT等等）去另一個地址——大大簡化換錢包或者合併資產嘅流程。智能錢包甚至可以允許你轉移錢包的「擁有權」：例如將個錢包賣比人或者轉交俾其他人（而EOA通常做唔到，因為EOA係死扣住一條私人鑰匙，而呢條key你唔應該亂俾人）。
Programmable Restrictions: You can impose arbitrary policies on your account usage. For instance, you might set a daily spending limit. If transactions exceed that in sum, the wallet pauses further transfers until the next day or requires an additional confirmation from you. This kind of rate-limiting can prevent loss of all funds if a key is silently compromised – the thief could only take e.g. 1% of your funds per day, giving you time to notice and react. Accounts could also restrict certain transaction types (e.g., “not allowed to call risky DeFi contract X unless an extra key signs”). This is analogous to how credit cards allow you to set limits or alerts on transactions over a certain amount.
可編程限制： 你可以為賬戶設定任何你想要嘅政策。例如設定每日消費上限：如果當日累積交易超過咗你設嘅上限，錢包可以自動暫停轉帳，等到第二日先恢復，又或者要求你額外確認先可以繼續。呢種rate-limit可以防止，如果你啲鑰匙被人偷咗都唔會一次過畀人掏空資產——小偷每日最多攞到你1%嘅資金，咁你有時間發現異常同補救。賬戶仲可以限制某啲類型嘅交易（例如：「唔准叫某啲高風險DeFi合約，除非多條key簽名」）。同用信用卡設置消費額和警報有啲類似。

In a nutshell, account abstraction grants unprecedented flexibility. A commentary from blockchain developers summarized it well: with EOAs users are “stuck with transactions that can’t be customized or automated – each one has to be signed individually.” But with account abstraction, “the game changes” and users can “set up recurring payments and dive into other forms of automation,” even greenlighting multiple actions in one go. It’s like moving from a manual stick shift car to an intelligent self-driving car that you can program with routes and rules – you go from doing every little action yourself to defining what you want and letting the system handle the mechanics.

總括而言，AA（賬戶抽象化）帶嚟前所未有嘅靈活性。有區塊鏈開發者咁形容：用EOA的話，「所有交易都唔可以定制或者自動化——每一單都要逐個簽名」。但用AA，「成個玩法唔同晒」，你可以「設置自動定期付款，甚至一個動作批准多個操作，體驗完全升級」。就好似你由自己打手波板車，升級到智能自駕車，你只需要講你個目的地同規則，系統就會自動處理點行。

Gas Fee Flexibility and Sponsorship

Gas Fee 靈活性及贊助

Another major benefit is the flexibility around gas fees that account abstraction affords. In current Ethereum, you must pay gas fees in ETH from your own account for every transaction. That’s a non-starter for many user-friendly experiences – imagine if every time you swiped your credit card, you had to also carry a second currency to pay a fee, and if you didn’t have it, the payment fails. That’s essentially the case with EOAs and ETH for gas.

AA另一個好處就係Gas費用可以靈活處理。依家以太坊所有交易都要求用自己賬戶用ETH去付gas。呢點對新手或者一般用戶嚟講，體驗好唔友善——諗吓，每次刷信用卡都要掏多件外幣俾手續費，如果冇就刷唔到；EOA同ETH嘅Gas就係咁嘅狀況。

Account abstraction breaks this constraint by enabling gas abstraction:

AA可以打破呢個限制，實現gas的抽象：

Your account (smart wallet) could be set up to pay gas using any token you have. For example, if you only hold USDC stablecoin, the wallet’s logic (in tandem with a paymaster or dex integration) could automatically convert a bit of USDC or use it to pay the miner/validator, so you don’t need ETH at all.
你嘅賬戶（智能錢包）可以設置用任何你有嘅代幣來付gas費。例如，你淨係有USDC，個錢包（可能結合支付合約或DEX）可以自動用USDC轉換或直接支付畀礦工/驗證節點，完全唔使準備ETH。
A sponsor (paymaster) can cover your gas. This opens the door to gasless transactions for users. A dApp might decide to pay for its users’ transaction fees to boost adoption – akin to a business covering shipping costs for customers as an incentive. This has been possible in limited ways via meta-transactions in the past, but account abstraction standardizes it and makes it more secure. Users can interact with a blockchain application without even realizing gas exists; the experience can feel like a free web2 app where operations “just work”. For instance, new users signing up could get a few free transactions sponsored by the app, smoothing their first-time experience.
**贊助商（Paymaster）**可以幫你出Gas。即係用戶可以享受「免Gas交易」。DApp可以選擇幫新用戶出費用，吸引更多人玩，好似商店幫你包郵費咁。有啲meta-transaction以前都做到類似，但AA將呢樣功能標準化，同時更安全。用戶可以用去DApp時連Gas呢回事都唔知，web2用慣嘅「一撳即得」體驗咁。例如新手註冊時，App可以送幾單免費交易，降低入門障礙。
Flexible fee logic: Maybe you want to automatically use the cheapest asset you have at the moment to pay fees, or dynamically choose between paying in ETH vs. another token depending on market rates – all that logic could be embedded in your wallet contract or a paymaster policy.
自訂收費邏輯： 想自動用當時手上最平嘅資產去付費，或者根據市價動態揀用ETH定其他token，都可以寫入錢包合約／Paymaster條policy度。

The ERC-4337 spec explicitly considers this a key feature: thanks to Paymasters, users “aren’t obliged to hold the native ETH token to engage with the network anymore. It’s a significant improvement for new users” entering Web3. And Rumble Fish’s analysis highlighted that with AA, dApps or even other people can chip in to pay someone’s gas as a gift or promotion, making onboarding much smoother. We’ve already seen organizations like Visa experiment with using account abstraction to let users pay gas with a credit card or via a third party, effectively making the crypto transaction feel like a normal online purchase. This kind of UX would be a giant leap in making blockchain applications mainstream-friendly.

ERC-4337標準明確話Paymaster係重點功能之一：「用戶唔洗再迫住hold住ETH才能玩網絡，新手會易用好多」。Rumble Fish嘅分析講到，AA後，DApp甚至第三方都可以幫人支付Gas，例如作為禮物或推廣，令新手上手體驗通順。有啲公司好似Visa，都試過用AA + 信用卡／第三方，幫用戶出Gas，變成好似網購咁簡單。呢啲用戶體驗係推動blockchain普及化的一大步。

Future-Proofing and New Possibilities

未來適應力和新可能性

Finally, account abstraction is important not only for what it enables today, but for how it future-proofs blockchain accounts for new technology and unlocks entirely new classes of applications:

AA唔單止解決咗現時問題，仲可以令blockchain賬戶未來upgrade無斷層，開發全新用例：

Post-quantum cryptography: Today’s Ethereum signatures (ECDSA) could be broken by quantum computers in the future. With account abstraction, we can gradually migrate to quantum-resistant signature schemes account by account, rather than needing a hard fork that changes how all signatures work. In fact, AA allows multiple signature schemes to coexist – some accounts could use traditional keys, others could require quantum-safe signatures like Lamport or BLISS, etc. Ethereum’s 4337 is seen as “one of the first steps towards creating quantum-computer resistant transactions” because it decouples the account verification from a fixed algorithm.
抗量子密碼學： 現時以太坊用的ECDSA簽名，未來可能會俾量子電腦破解。有AA，大家賬戶可以逐個逐個慢慢換去量子防護的簽名，不用一次過hard fork全系統。AA仲支援多種簽名共存——有啲賬戶用舊key，有啲用Lamport、BLISS呢啲量子安全方案。ERC-4337都被視為「走出抗量子的一步」，因為驗證邏輯唔再死扣一套算法。
Role-based access and modularity: Accounts can be programmed for role-based access control. For example, you could designate a “trading key” that’s only allowed to make trades but not withdrawals, or a “deployer key” that can deploy contracts but not move funds, all under one account contract. This is useful for organizations or power users who want fine-grained control.
角色權限模組化： 賬戶可以自訂角色和權限。例如設一條key只准交易不准提現、另一條key只准發合約唔准動資金——全部都係同一個賬戶底下。對公司或者高端用戶要精細控制好有用。
First-class multisig and shared accounts: Account abstraction could make multi-owner accounts a first-class citizen across the ecosystem. This means dApps and protocols can natively support interactions with multisig accounts more easily. It also means team or family wallets become easier – an account contract can be owned by N people, each with certain rights, which is not straightforward with EOAs. In fact, the reference text noted account abstraction enabling “team wallets” as a use case, referring to wallets that multiple people manage with programmed governance rules (ideal for business treasuries, DAO funds, etc.).
多簽&團隊共同賬戶： AA可以令多人的共管賬戶成為生態一等公民。DApp／協議層可以原生兼容這類帳戶。即係公司、團隊、家庭用途都方便咗好多——同一賬戶由N個人合管，每人分不同權限（EOA就難做到）。參考文件有提過「團隊錢包」就係AA的直接用例——即多個人有程序化治理去管共同資金（適合企業財庫、DAO資金等）。
On-chain identity and reputation: Since account contracts can contain logic, they could integrate things like reputation scores or allowlists for DeFi (e.g., an account that only allows interactions with whitelisted protocols for safety until the user changes settings). They might also integrate with identity systems, like requiring a certain credential or NFT to unlock some features. This blends into the realm of smart accounts serving as both wallet and identity hubs.
鏈上身份和信譽： Account contract可以寫邏輯，例如連接信譽分、DeFi白名單（如賬戶預設只准對白名單協議操作，保障安全）、仲可以和身份系統整合（例如指定NFT才解鎖某功能）。AA錢包於是可以同時做到錢包+身份認證站。

All told, the benefits of account abstraction span security, usability, flexibility, and future-readiness. It’s about making crypto accounts as powerful and convenient as modern software will allow, without sacrificing the self-custody and decentralization principles. No wonder that many in the Ethereum community see it as pivotal for driving the next wave of adoption. As one source put it, account abstraction is widely viewed as an “important stepping stone in Ethereum’s path to massive user adoption.”

總括而言，AA在安全、易用性、靈活度、未來upgrade等多方面都全面提升。最重要係令crypto賬戶既有自主權又享受到現代軟件的高效和易用。怪不得咁多以太坊圈內人認AA係推動下一波普及的關鍵路程。有開發者甚至講：Account Abstraction係以太坊去到主流化「必經的重要一步」。

Having painted the picture of what AA enables, let’s ground this discussion by looking at some real-world implementations and examples of account abstraction in action, as well as what’s happening right now in the ecosystem.

講到咁多AA的好處，不如實際睇下而家有咩真實應用例子同執行個案，睇下生態係點發展緊。

Real-World Applications and Examples

真實應用及例子

Account abstraction may sound theoretical, but it’s already being implemented and tested in the wild. Here are a few notable examples and scenarios that illustrate its impact:

AA聽落似係理論，但實際已經有唔少落地實踐同測試。以下係幾個有影響力的例子同應用場景：

Smart Contract Wallets (Social Recovery & Multisig): Projects like Argent wallet have been early pioneers of smart contract wallets that offer social recovery and trusted contacts. Argent’s wallet (even before ERC-4337) let users nominate “guardians” to help restore access if they lost their key – this was achieved via a custom contract for each user. With ERC-4337 now live, such wallets can plug into a standardized infrastructure and potentially become more common across the industry. Similarly, Gnosis Safe (now called Safe) has been a widely used multi-signature wallet (mostly for teams/DAOs). Safe is essentially an account abstraction use-case (multiple owners controlling one contract account). In fact, the Safe team is actively embracing AA; they even developed a prototype to leverage ERC-4337 and are looking at how upcoming protocol changes (like EIP-7702) will support migrating existing Safe accounts to first-class smart accounts. These examples show that enhanced security wallets for both individuals and organizations are a clear immediate win for AA.
智能合約錢包（社交恢復、多簽）： 例如Argent智能錢包好早期就做到social recovery（社交恢復功能）同trusted contacts。咩意思？Argent（未有ERC-4337都做到）用戶可以揀「守護者」幫你搵返賬戶（如果你唔見咗key），每個用戶都有一個自訂合約支援。而家ERC-4337標準推咗出嚟，呢類智能錢包可以直接連結標準化基礎設施，更易成為行業主流。另一例子，Gnosis Safe（而家叫Safe），多年來都係大型團隊/DAO用嘅多簽錢包。Safe本身就係AA用例（多人控一個合約戶口）。Safe團隊積極擁抱AA，測試咗ERC-4337 prototype，研究點樣用未來如EIP-7702標準，將Safe賬戶轉做智能合約戶口。呢啲例子說明，AA對提升個人同公司安全嘅錢包即用即有突破。
Gas Sponsorship by DApps: We’re seeing decentralized applications experiment with covering user gas fees to improve UX. For instance, a blockchain game or a decentralized exchange might use a Paymaster (per ERC-4337) to let users trade without holding ETH for gas – the dApp sponsors the gas, perhaps recouping the cost via a slightly higher protocol fee or as a marketing expense. This gasless transaction experience is very attractive for onboarding new users. A DeFi platform could literally advertise “no ETH needed to start lending – deposit using USDC directly”. Some wallet SDK providers (like Biconomy or OpenGSN) have provided meta-transaction frameworks historically; with account abstraction, this can be done more natively and securely. In one case, an Ethereum foundation supported project demonstrated a system where users could even use a credit card to indirectly pay fees – the Visa
DApp贊助Gas費： 有唔少去中心化App已開始幫用戶出Gas，為提升UX做實驗。例如區塊鏈遊戲或者DEX交易所利用ERC-4337 Paymaster，俾用戶買賣時都唔洗準備ETH——DApp直接幫你俾Gas，可能自已加少少服務費或者當宣傳。免Gas初心者好有吸引力。某DeFi平台甚至直接標榜「無需ETH，USDC直接存款做借貸」。有啲wallet SDK供應商（如Biconomy、OpenGSN）從前做meta-transaction，而家AA之下可以原生更安全實現。一個由以太坊基金會支持的項目甚至做咗用信用卡間接出Gas，Visa…research we mentioned allowed a wallet to pay gas by charging a Visa card, all mediated by the wallet’s logic and a paymaster. While charging a credit card for gas isn’t common yet, the fact it’s possible highlights how far we can abstract the blockchain mechanics away from the user.

我們之前提到有個研究，讓錢包可以用 Visa 卡付款來支付 gas 費，整個過程由錢包邏輯同 paymaster 中介處理。雖然而家用信用卡畀 gas 費未係主流，不過事實係做到，反映咗我哋可以幾徹底咁將區塊鏈底層嘅操作抽象化，等用戶根本唔使理。

Recurring Payments and Subscriptions: The concept of automated recurring payments from a self-custodial wallet was practically unheard of before, because an EOA can’t initiate a payment on its own at a future date. With account abstraction, however, auto-payments become feasible. The Visa proof-of-concept on StarkNet is a prime example: they used account abstraction to implement a pull-based payment (the biller could trigger the payment from the user’s wallet on the due date, because the wallet had pre-authorized it). Another hypothetical example: a streaming service could deploy a smart contract that, each month, pings your wallet contract for the subscription fee; your wallet’s code could verify it’s the legitimate service and automatically pay them in, say, a stablecoin – all without you signing in every month. This kind of convenience was typically missing in Web3, potentially forcing users into custodial solutions if they wanted such features. Account abstraction brings it to self-custody.
定期付款同訂閱服務： 自我保管錢包（self-custodial wallet）做自動定期付款呢個概念，以前基本上無人諗過，因為 EOA（傳統帳戶）本身係唔可以自己揀幾時送錢俾人。但有咗帳戶抽象（account abstraction）之後，自動扣數就變得可行。Visa 喺 StarkNet 嘅概念驗證就係好好例子：佢哋用 account abstraction 做咗「拉式支付」（即收款方可以喺預設嘅日子主動向你個錢包撳過數，因為你預先 authorize 咗）。再舉個例：串流服務可以 deploy 智能合約，每個月問你個錢包收訂閱費；你個錢包啲 code 可以 check 服務商身份啱唔啱，然後就自動用穩定幣幫你畀錢—你都唔使每個月自己登入。以前 Web3 資歷通常無呢啲咁方便嘅功能，要有就要用託管錢包。account abstraction 令自我保管都做到呢類便利。
“One-Click” Experiences & Composability: Consider an NFT marketplace where buying an NFT might involve multiple steps (approve token, then trade, etc.), or a DAO participation that requires locking tokens then casting a vote. With AA wallets, projects can design flows where the user does a one-click “buy” or “participate” and behind the scenes the wallet contract bundles the necessary steps. We already see this with some DeFi aggregators that do meta-transactions, but with native AA it could be more prevalent and simpler to integrate. This increases composability of dApps – your smart account could interact with multiple protocols in one go, which encourages developers to create richer features without worrying that users will drop off after the first of several transactions.
「一按即完成」體驗同可組合性： 諗下 NFT marketplace，買一件 NFT 要 approve token、做交易等，步驟唔止一樣。又或者參加 DAO，要 lock token，再投票。有了 AA 錢包（account abstraction wallet）之後，開發者可以設計一條龍流程，用戶只要「一撳」，背後個錢包合約自動幫你 bundle 埋所有細步。依家部分 DeFi aggregator 透過 meta-transactions 都做到類似效果，但係如果原生支持 AA，將會更盛行同更易整合。咁樣 dApp 間可以更加靈活—你個 smart account 一個操作已經可以同唔同協議互動，開發者就唔怕用户喺中途會流失，可以放心設計更豐富功能。
Layer-2 Adoption and Cross-Chain UX: On Ethereum Layer-2 networks like StarkNet and zkSync (which have native AA), users are getting a taste of these benefits from day one. A user on StarkNet, for example, might create their account by deploying a contract (there is an initial one-time cost to deploy your account contract) and thereafter enjoy features like choosing any token to pay fees. As these L2s gain users, the expectation for such convenience will grow, pressuring other chains to adopt similar ideas. Moreover, account abstraction can help with cross-chain experiences. Some in the community talk about “chain abstraction” hand-in-hand with account abstraction. For instance, a smart wallet could abstract which chain an operation happens on – you could initiate an action and the wallet (via relays or bridges) handles getting it executed on the appropriate chain, returning the result to you, without you manually switching networks or holding multiple chain tokens. This is still early-stage, but conceptually a smart account could manage resources on multiple chains if designed to, giving a unified UX.
Layer-2 應用同跨鏈用戶體驗： 以 StarkNet、zkSync 等以太坊 Layer-2 網絡為例，佢哋原生支援 AA，用户一開始已經感受到呢啲新便利。例如 StarkNet 上，開戶就係 deploy 一份合約（deploy 帳戶合約有一次性成本），之後可以揀用咩 token 畀費用。當 L2 有多啲人用，大家對便利功能會有期望，其它鏈都要諗方法跟進。account abstraction 仲幫到手做跨鏈操作。社群有啲人會一齊講「鏈抽象」同 account abstraction。例如 smart wallet 可以抽象化「邊條鏈執行指令」—你發動時，個錢包（透過橋、代辦等）自動揾最啱鏈做，然後回返結果唔使你自己 switch network 或揸好多鏈 token。呢個概念仲係初步，但理論上 smart account 可以分布多條鏈管理資源，提供一體化用戶體驗。
Developer Tooling and New Services: A host of new services are popping up to support account abstraction. For example, providers offering Wallet-as-a-Service (WaaS) that handle the deployment of smart wallets for users and manage keys in user-friendly ways (some integrate secure enclaves in phones or cloud backups, etc.). While we won’t promote specific companies, it’s notable that many startups and projects are actively building AA tooling – from SDKs that let any dApp spin up an AA wallet for their users, to specialized paymasters that handle gas conversions. This means the ecosystem is rapidly moving towards making AA seamless. As these tools mature, more apps can adopt AA without reinventing the wheel, and users might use AA without even knowing it (for example, a game might automatically give each user a contract wallet in the background linked to their email login – the user just knows they have a game account, which under the hood is a smart contract wallet tied to their email-authenticated key).
開發者工具同新服務： 市場湧現咗好多 account abstraction 相關新服務。例如有啲供應商提供 Wallet-as-a-Service（WaaS），幫用戶部署智能錢包，又有用戶易用嘅鑰匙管理方式（有啲甚至會用電話安全芯片/安全區，或者上雲端備份）。唔特別推介啲公司，但可以見到好多初創積極開發 AA 工具，由一個 SDK 借任何 dApp 幫用戶開 AA 錢包，到專門做 gas 轉換嘅 paymaster 應有盡有。生態好快就做到 AA「無縫」支援。當啲工具成熟多啲，app 開發者都唔使自己再生發明，可以直接整合 AA，用戶甚至唔覺就已經用咗（例如一隻 game 背後自動開個合約錢包，用 email 登入自動連結—用戶做咗個 game account，但實質已經係合約錢包配對佢 email auth key）。

All these examples reinforce that account abstraction isn’t just a theoretical upgrade; it’s happening now across various fronts, bringing concrete improvements. However, it’s not all sunshine and roses yet. Like any new technology, there are challenges and trade-offs to be aware of. It’s important to examine these to get a balanced view.

呢啲例子證明 account abstraction 唔淨係理論升級，而家唔同範疇都已經有明顯應用同改善。但都唔係完全冇問題，好似所有新科技一樣，帳戶抽象都有挑戰同要取捨位。值得仔細睇下啲利弊，先有客觀平衡觀點。

Challenges and Limitations of Account Abstraction

帳戶抽象（Account Abstraction）嘅挑戰同限制

While account abstraction opens exciting possibilities, it also introduces new complexities and considerations. Here are some of the challenges and limitations to keep in mind:

帳戶抽象帶來好多新可能，但同時都會有新複雜性同要注意嘅地方。要小心水下面啲挑戰同局限：

Smart Contract Risk: By turning user wallets into smart contracts, we inherently introduce smart contract risk to personal accounts. A bug in the wallet’s code could be disastrous – for instance, a flaw could allow an attacker to bypass security or drain funds. With EOAs, the “code” involved in your account is basically just ECDSA signature verification, which is a well-tested cryptographic primitive. Smart wallets are far more complex. Although the core AA frameworks (like the EntryPoint contract in ERC-4337) are audited, the security of each wallet implementation can vary. As one developer guide noted, when using an AA wallet, you’re “deploying an immutable contract” and if a bug is found, it can be challenging to patch since that contract code can’t be easily changed. Some wallet contracts might include upgradeability or migration features to mitigate this, but that then introduces trust considerations (who can upgrade it?). Diligence in auditing wallet contracts is crucial.
智能合約風險： 由於用戶錢包變成智能合約，自己帳戶自然多咗智能合約風險。錢包 code 出錯分分鐘好嚴重—例如有啲漏洞就俾黑客繞過安全措施或者攞走晒錢。EOA 其實只係做 ECDSA 簽名驗證，好成熟穩陣。智能錢包複雜好多。雖然 AA 核心框架（例如 ERC-4337 嘅 EntryPoint）有做審計，但每個錢包實際部署嘅安全度可以唔同。開發者手冊亦提過，用 AA 錢包即係「deploy 一份唔可以改既合約」，即使發現 bug，要 patch 都幾麻煩，因為 code 唔易轉。有啲錢包可能加咗 upgrade 機制或者 migrate 功能處理，但又會涉及信任（邊個可以幫你 upgrade？）。所以錢包審計一定要夠嚴謹。
Complexity and New Failure Modes: The AA architecture (with bundlers, paymasters, separate mempool) is more complex than the status quo. This means more components that could fail or be attacked. For example, what if the bundler network is not sufficiently decentralized early on? Could bundlers censor certain UserOps or demand high fees? There’s a risk of centralization if only a few actors become dominant bundlers. Over time, it’s expected that many Ethereum nodes or miners/validators themselves might run bundler software (especially if economic incentives are there), but in early stages, users are trusting that the mempool of UserOps and bundlers are working honestly. The EntryPoint contract is another central trust point – if a vulnerability were found there, it could affect all AA users. The Ethereum community has taken precautions (the EntryPoint can be replaced via an update mechanism if a bug is found, under a multisig governance by devs until full decentralization), but it’s a key piece to watch.
複雜程度升高、潛在失效模式： AA 架構（有 bundler、paymaster、獨立 mempool）比現時傳統 approach 複雜啲，即係多咗啲會 fail 或者俾人攻擊嘅地方。例如如果早期 bundler network 未夠去中心化，會唔會出現 censorship（選擇性封鎖）或者亂收高 fee？如果只有幾個大 bundler，返轉頭又會中央化。長遠可能多啲節點或者 validator 自己跑 bundler，但初期大家都要信 userOp mempool 同啲 bundler 夠唔夠公平老實。EntryPoint contract 亦都係重要信任點—入面有漏洞就會影響晒所有 AA 用户。以太坊社群小心處理—EntryPoint 有得以多簽 upgrade，如果爆 bug 都可以換新，長遠再落去才完全去中心化—總之係個焦點。
Resource Costs (Gas and Deployment): Using a smart contract wallet has overhead. There is a one-time deployment cost to create your account (you have to publish a new contract on-chain for each user wallet, unless using a counterfactual deployment pattern where it’s created at first use). This could cost a few dollars in gas on Ethereum mainnet, which might deter some users or require wallets to sponsor that. Additionally, each operation through a smart wallet might be slightly more expensive in gas than a simple EOA transaction because it involves calling the EntryPoint, executing additional code, etc. However, some of this can be offset by batched execution efficiencies. Still, for heavy on-chain activity, those costs add up. This means, at least initially, account abstraction might be more common on Layer-2s (where gas is cheaper) and only for higher-value use cases on Layer-1. The good news is that Ethereum developers are aware of this and are working on protocol changes to make AA more gas-efficient. For example, proposals like “InitCode compression” or other EIPs aim to reduce the cost of deploying and using smart accounts, and in the long run if AA becomes the default, the protocol can optimize for it.
資源成本（Gas 同部署）： 用智能合約錢包會有額外成本。開戶要 deploy 合約（除非用 counterfactual 部署，首次用先真係部署），主網 gas 價可以幾蚊美金，嚇走部份用戶或者叫錢包自己 sponsor。加上每次操作都可能比 EOA 用多啲 gas，因為又要經 EntryPoint 又要 execute 多啲 code。不過，打包執行某啲操作可以 offset 少少，但大量 on-chain 用戶成本都會計埋。即係一開始 account abstraction 可能只會喺 L2（gas 平啲）較普及，L1 只用嚟做高價值用途。好消息係 ETH Core Developer 已經積極搵方法減低 AA gas 成本，例如「InitCode 壓縮」等新提案，長遠成為標配之後，協議級可以好針對性咁優化 support。
Key Management is Still Key (Literally): It’s important to note that account abstraction doesn’t eliminate private keys – it just adds layers around how keys are used. You still ultimately need some form of private key or secret to authenticate as the owner of an account (even if that key is split among multiple parties or stored in hardware, etc.). If a user chooses poor security for their keys, they can still get hacked. AA provides tools like social recovery, but users must actually use them and set them up properly. Some critics point out that many users might stick to default settings, which could be a single key controlling the account contract (basically replicating an EOA, but with more complexity). In such cases, if they never configure guardians or 2FA, they haven’t gained much security – and they might even be at more risk if they don’t understand the new wallet model. In summary, account abstraction greatly improves potential security, but does not guarantee it. Users will need good UX to guide them to safer setups (e.g., prompts to add a guardian or a backup key during wallet onboarding).
密鑰管理依然關鍵： 要留意 account abstraction 冇咗私鑰都唔得，只係包咗多幾層（一樣要有啲秘密—for example, hardware 硬件儲存、多人分散喺手）。用戶安全做得唔好，都一樣有機會畀人 hack。AA 提供咗社交恢復等工具，但都要用戶自己開通同設定。有人批評大部分人會黐住預設設置，即係一條 key 控制晒個 account contract（即係 EOA 嘅安全加複雜度）。如果唔搞 guardian/二次認證反而未加到什麼保護，甚至唔明錢包新 model 有機會更危險。簡單講，AA 係令你容易設到高安全度，但唔包一定安全，用戶 UX 上要有提示等（例如 onboarding 時建議/提醒加 guardian 或後備鑰匙）。
Not Yet Universal: As of 2025, account abstraction via ERC-4337 is available on Ethereum, but it requires wallet providers to support it. If your current wallet (say MetaMask or hardware wallets) doesn’t support creating and managing 4337 smart accounts, you can’t benefit from AA without switching. We are in a transition period where both EOAs and AA accounts coexist. This can cause user confusion and friction. For instance, AA accounts have their own address (which looks like any Ethereum address, but it’s actually a contract). If someone sends ETH to your AA wallet address, that’s fine – it’s an address – but to send ETH out, you’ll be going through the AA flow rather than a simple EOA transaction. Power users might worry about compatibility: “Will this dApp support my
未普及化： 去到 2025 年，ERC-4337 嘅帳戶抽象已經喺以太坊用得，但要你個錢包服務商有支援。如果你用緊嘅錢包（好似 MetaMask 或硬件錢包）仲唔支援 4337 smart accounts，就用唔到 AA 功能，非要換錢包。依家正值 EOA 同 AA account 共存過渡期，用戶容易混淆同卡住。例如 AA account 有自己地址（睇落似普通以太坊地址，但其實係合約來的）。有人 send ETH 入你 AA 錢包無問題—都係 address—但 send ETH 出就要用 AA flow，就唔係傳統 EOA 簡單交易。進階用戶會擔心：呢個 dApp 支唔支援我......smart wallet？”一般嚟講，如果個 AA 錢包設計得好，其實應該可以同所有 dApp 一齊用（因為對 dApp 嚟講，其實淨係見到有個地址去執行 contract call）。但有啲較底層嘅工具（例如某啲區塊鏈瀏覽器或者舊式錢包）可能未完全識別到呢啲交易。隨住時間過去，好似 ERC-4337 呢啲標準應該會愈嚟愈 seamless，不過整個生態圈仲要追上——例如鏈上瀏覽器、硬件錢包 firmware 等功能，可能都要更新先可以正確顯示同簽署 AA 操作。
互通性同多鏈支援: 如果你要喺多條鏈（L1、L2、sidechains）都用你個 smart account，你可能要喺每條鏈都部署一次 contract，幾麻煩。有啲人而家正努力研究點樣可以做到 account 部署可以「複製」去唔同鏈，咁你就可以保住同一個地址同功能。不過，未完全解決之前，要記住你喺一條鏈用 AA，唔等如自動可以喺另一條鏈用——每條鏈都可能要做一次 setup。
現有用家過渡難題: 而家已經有幾百萬個 EOA 㗎啦，當中有啲仲擺咗好貴重嘅資產（例如綁死或唔可轉讓嘅 NFT，即係你唔可以簡單搬去新 smart wallet）。如果用戶想轉用 Account Abstraction，應該點做？有一種方法係靠協議升級，好似 Vitalik 同其他人提議嘅 EIP-7702，可以令 EOA 「領養」 smart contract 功能而唔洗轉地址（下一段詳述）。但未升級之前，用戶可能要開新帳戶至有 AA 功能，對於有啲資產既定咗喺舊地址會幾麻煩。再加上教育問題：用戶要明點解要轉去 smart wallet，同要克服「冇壞就唔好搞」嘅心態。社區要宣傳佢啲好處（又或者錢包產品會出 one-click 遷移工具，等大家易過戶）。

雖然有呢啲挑戰，以太坊社區普遍都覺得 AA 嘅優點遠多過缺點，而家大部份限制都積極解決緊。所有新基礎技術初期都會好複雜，但之後會愈嚟愈順。好似第一批智能手機又慢又唔耐用，但而家全部都離唔開、好易用咁；AA 錢包而家可能新鮮又唔慣，但幾年後無社交恢復、免手續費交易等功能你可能完全唔想用返舊式錢包。

帶住呢個平衡嘅觀點，我哋可以展望下 Account Abstraction 未來嘅方向。佢仲會點發展？未來想令 smart account 成為新常態，有乜新搞作？

全面 Account Abstraction 之路

Account Abstraction 喺以太坊（特別係用 ERC-4337）當然係大 milestone，但好多核心開發者都話，呢個純粹係一個旅程中間嘅一步，唔係終點。終極目標係做到「全面 account abstraction」，即係底層協議層面全部帳戶都係 smart account，EOA 慢慢消失。要達到呢個位，未來幾年應該會有進一步升級同仔細過渡策略。以下係未來展望：

1. 協議級整合:
而家 ERC-4337 係應用層運作，靠 EntryPoint smart contract 用番以太坊現有交易系統做後台。長遠嚟講，以太坊可以直接喺協議（Layer 1）層面整個 AA 入去，令流程簡化。例如加一種新 transaction type，或者修改共識規則，等 smart contract wallet 唔使再靠 bundler 都可以直接發交易。早期好似 EIP-2938（雖然冇落實）都有提議過協議級「AA 交易」型態。社區選擇咗 4337 掃條路，之後都唔排除會有更深層改動。如果原生整合 AA，以太坊就唔使分開一個 UserOps mempool，有機會降低 gas 成本（因為 EntryPoint 啲運作可以由協議層優化）。近期仲有個「統一 mempool」構思（草案叫 RIP-7560），即係將 UserOps 同一般交易合併埋同一個池，執行更平。雖然聽落技術性多，但意思就係架構簡單化，成本下降。

2. EOA 過渡去 Smart Account:
為咗做到 fully account abstraction，以後可能唔會再新開 EOA。所有新 account 開始都自動係 smart account。如果錢包軟件（例如 MetaMask）直接只開 4337 smart wallet 而唔係純 key account，user 其實唔覺意都已經用咗 contract account。最難係點轉現有 EOA。有個叫 EIP-7702 提議，有機會 2025「Pectra」網絡升級時實行。EIP-7702 令 EOA 可以直接喺自己個地址執行 smart contract code。佢原理係 EOA 可以指一個「委託」合約——即係掛件 smart logic 喺個 EOA 上面。有人同 EOA 交互時，網路會執行連接咗嘅合約 code（好似 delegatecall），即係當 EOA 係 contract 咁處理。即使原本係 EOA，都即時有齊 smart account 功能，份 key 仍然控制到。實際即係如果你舊有 EOA 攞嚟收重要嘢，可以加 smart contract 模組——例如 multi-sig、social recovery——又唔使轉地址，唔使搬資產。好似幫舊架車加個現代自動導航引擎一樣。

3. 淘汰「單一私鑰」控制:
要達到真正 account abstraction，以太坊有機會規定轉咗 smart account 嘅戶口，之後就唔准用返原本個 key 直接 sign 交易（即係唔可以繞過 smart logic）。有個有趣提議叫 EIP-3607，建議只要帳戶有 contract code，就 reject 所有普通交易（即係 presumably 用返舊 key 嘅交易）。換句話講，如果你 convert 咗 EOA 做 smart account，就唔會走回頭——合約已經攞正牌控制，單憑私鑰都唔郁得資產出 contract 條 rules。呢類改動一定要細心規劃，唔可以搞到用戶窒咗，不過呢個就係「無 EOA」嘅最終狀態。平行之下，好似 EIP-1271（定義 contract-based signature verification 標準）已經普及，contracts 可以去啲要「簽名」認證嘅 protocol 行使用戶身份。長遠嘅 idea，係而家只認 EOA 簽名嘅地方，都接納 smart account base 嘅算法，到時 contract account 就正式成為鏈上 full-featured account。

4. 增強預設功能:
喺 fully account abstraction 環境，新開嘅 wallet 有機會預設已經搞好晒啲而家得 optional 先有功能。例如新帳戶開戶時可能自動 prompt 你設好 social recovery（可能要揀幾個 guardian contact）。這些恢復選項唔再係事後加，係一開始就有。同樣，multi-device、多重身份驗證（multi-factor）都可以一開始就預先整好。好似剛剛講，multi-signature 格式可以融合：你個戶口允許用電話指紋（可能連 Secure Enclave ）做主要認證，再加 hardware key 做後備。Account abstraction 容許全部合一個帳戶。實際上令錢包更易用（唔洗死記 seed phrase），同時又更 resist 新型威脅，例如量子計算。

5. App 廣泛支援同新場景:
隨住愈來愈多 dApp 同用戶採用 AA，預咗會見到多咗好多創新玩法。例如 DeFi 平台可能對 AA 錢包有啲獨特功能（例如「用 AA 錢包，可以直接喺我們 UI 落限價單」）。區塊鏈遊戲搞少啲簽名麻煩。DAO 管理都有新 idea——個 account 可以內建投票規則，自動 vote 或 delegate。可能性無限，基本上所有大家以前想有但 EOA 做唔到嘅用戶層邏輯都會開始變做標配。

6. 其他區塊鏈都仿效: 以太坊帶頭，但其他 chain 都留意住呢個方向。有啲自己整啱自己嘅 account abstraction。講咗 StarkNet 同 zkSync，也有啲新網絡（Polkadot、Cosmos）研究「smart key」或彈性帳戶。例如 Polkadot 可以原生設定社交恢復，拉幾個朋友帳戶做後備。見到以太坊 AA 得，預咗都會廣傳去唔同 platform——個核心思想係帳戶安全、執行邏輯 user 可以自由設計。發展下去，「account abstraction」呢個 term 可能幾年後都少講，只會叫 smart account 或 account。

總結，account abstraction 嘅未來係希望所有 crypto 帳戶，一開波就有 smart contract 咁強大嘅功能，而舊有 EOA 會慢慢淡出。要去到呢步，要靠一系列細緻升級（例如 EIP-7702 同其後），同埋 user-friendly 遷移路徑。成個過程需要好細心——你唔可以force everyone to switch overnight – but the momentum is there. Ethereum’s developers have sketched a roadmap where ultimately a majority of users transition to smart accounts, reaping benefits in security and usability, and then the protocol can optimize around those assumptions (like maybe one day, Ethereum could drop the concept of mandatory gas paid in ETH if everyone is using paymasters and such – but that’s speculative and far out).

Final thoughts

帳戶抽象喺區塊鏈帳戶管理方面帶嚟咗一個範式轉移。透過容許用戶用智能合約作為帳戶，打破咗以往僵硬嘅限制，開創咗一個將來——用加密貨幣可以同用傳統銀行app一樣易，甚至更加簡單，同時用戶仲可以掌控更多安全權限。唔再係一條鎖匙唔見就萬劫不復，亦唔需要次次手動簽名或者要留定啲ETH淨係為咗用dApp。依家有咗帳戶抽象，啲功能例如社交恢復、多簽安全、自動付款、批量交易、甚至免GAS使用，都唔再係咩偏門嘢或者夢想——而係逐步成為基本工具箱入面嘅標準。

實際嚟講，帳戶抽象之所以重要，係因為佢直接針對緊擴大加密貨幣用戶群嘅兩大最大障礙：用戶體驗同安全。佢帶嚟靈活性（可以自訂錢包規則、自選認證方式）同包容性（例如可以畀其他人幫你付費、用簡單登入方式、遇錯誤可以復原），但又唔犧牲非託管（non-custodial）嘅精神。呢個技術係Web3變得易用嘅基石。事實上，Ethereum領導層同好多圈內人都覺得帳戶抽象係生態成敗關鍵——睇法係，如果唔解決現有錢包嘅爛用戶體驗同安全陷阱，加密貨幣係唔可能達至十億用戶。帳戶抽象正正係解決方案嘅大部份。

依家，我哋已經見到呢個新模式喺Ethereum上面（例如ERC-4337），同埋唔同Layer-2網路原生實現嘅早期階段。未來幾年，好大機會逐步有更多無縫整合——你用一個去中心化app時，可能完全唔覺你個“account”實際上係一個智能合約，幫你自動處理晒一切。錢包供應商、dApp開發者、最終用戶全部都受惠：阻力小咗，可能性大咗。

當然，喺採納新模式時，都要保持警覺——智能合約錢包必須被細心設計同審計，用戶都應該了解新功能，好似社交恢復等等。但呢啲都係可以管理嘅挑戰，遠遠唔及一個安全又方便嘅加密生態系統帶嚟嘅好處。

總結嚟講，帳戶抽象可以睇成區塊鏈技術成長嘅一大步。正如互聯網由以前command line進化到而家易用嘅網頁介面，區塊鏈都由舊時硬淨嘅私匙管理時代進化到智能帳戶年代。呢場革命雖然發生緊喺基建層，但用戶會直接感受到影響：資金更安全、登入更方便、玩digital asset有更多玩法。隨住技術進步，你唔好覺得驚訝，日後“忘記密碼”或者“授權app用24小時”呢啲汁汁，都可以係你crypto字典內出現——呢啲就係帳戶抽象發揮作用，令加密貨幣感覺同你用開其他數碼服務一樣自然，但又保留到最初吸引大家跳入區塊鏈嘅自由同主權。