KoinlyticsEthereum's original wallets are boring. One private key controls everything. Lose the key, lose the money. Every transaction needs ETH to pay gas. Nothing else is possible unless the wallet itself is a smart contract. That is what account abstraction fixes without a hard fork.
ERC-4337 in one paragraph
Instead of transactions from EOAs (externally owned accounts), users produce UserOperations. A bundler collects UserOperations and submits them in a wrapper transaction. An EntryPoint contract validates each op's signature, calls the user's smart account, and settles gas. From the wallet's point of view, gas is paid by a paymaster contract, which can subsidize it, take it in USDC, or pull it from a subscription.
What actually becomes possible
- Gasless transactions. A dApp pays gas for you.
- Pay gas in any token. Paymaster accepts USDC, DAI, whatever, converts under the hood.
- Session keys. A game gets a limited key that can only make in-game moves for 24 hours. No sign popup per turn.
- Social recovery. Lose your device? 3 of 5 friends can help you rotate keys.
- Multisig by default. Every account can require multiple signers without a Safe-style contract on top.
- Batch transactions. Approve + swap in one signature.
The trade-off
Every advantage above adds gas. A basic UserOp costs roughly 40-50% more gas than a plain EOA tx today. On mainnet that hurts. On L2s, where gas is a fraction of a cent, it is a rounding error. This is why 4337 usage is exploding on Base, Arbitrum, and Optimism first.
Where the standard is going
EIP-7702 (in Pectra) lets a regular EOA temporarily behave like a smart account for a single transaction. That closes the gap for the wallets that don't want to migrate. Longer term, EIP-3074 style delegation may replace 4337 entirely. The user-facing outcome stays the same: your wallet stops feeling like 2015.