Urgent.News

What's breaking now, across thousands of outlets.

Tech

Zero-Gas Autonomous Agents: Account Abstraction on Base L2

Zero-Gas Autonomous Agents: Account Abstraction on Base L2 Autonomous AI agents break the moment they touch an onchain transaction lifecycle. The standard stack usually fails like this: you spin up an LLM runner, wire it to an MPC or raw EOA private key, and fund it with $50 worth of ETH. Three days later, the agent stops dead. Gas spiked, nonce management desynchronized under concurrent…

Zero-gas autonomous agents on the Base L2 network present a novel approach to account abstraction, aimed at eliminating friction in onchain transactions. The architecture consists of three key components: a Smart Account (Kernel/Safe + 4337 module), a Paymaster Policy, and a Treasury Anchor. The Smart Account holds an ephemeral key for proposing operations without draining treasury assets.

The Paymaster Policy involves offchain services or onchain vaults that sponsor gas for transactions that meet specific validation rules. The Treasury Anchor ensures all realized value routes back to a secure treasury, maintaining a bare operational hot-wallet.

In this setup, transactions are submitted as UserOperations instead of raw EVM transactions. The bundler picks up the UserOperation, queries the Paymaster, and the Paymaster pays Base sequencer fees on behalf of the agent. Implementation can be achieved through a Python script using web3.py and eth_account libraries. The script builds a sponsored UserOperation on the Base network, requesting Paymaster sponsorship and injecting the response into the UserOperation.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in Tech

Information‑gain design trims training compute

Amortizing sequential Bayesian experimental design turns the expensive, inference‑time search for information gain into a cheap forward pass, letting LLMs ask smarter questions with far less compute.

  • Amortizing sequential Bayesian design reduces compute for LLMs.
  • Deep Adaptive Design learns policy for next query selection.
  • AR-ED policies improve information gain with less computational cost.

Voice Robot for Outbound Calls: A Latency Budget from ASR to TTS

A voice robot that confirms appointments can be perfectly accurate and still feel broken. The customer says "yes, I'm coming", then hears nothing for two seconds, says "hello?", and the robot starts…

  • Voice robot latency critical for seamless customer experience
  • Turn latency target of 700 ms or less for natural feel
  • Call path divided into SIP trunk, media server, VAD, ASR, dialog logic, TTS, and RTP stages

I Replaced MongoDB with Turso Without Touching Roughly 1,250 Database Calls

In October 2026, I moved the Future Thinkers Club platform off MongoDB and onto Turso. The interesting part is not the move itself. It is that the application code barely noticed: roughly 1,250 database calls stayed exactly as they were. Why move at all The platform runs on Vercel, on serverless functions.

  • Platform migrated from MongoDB to Turso without code changes
  • 1,250 database calls retained same functionality
  • Interface adapter enabled seamless database switch

More from Monday 9 November →