From Zero to Full App in Minutes with nora, the Coding Agent
summary
Nora runs inside a VS Code fork and combines foundational models (Claude, Gemini) with post-training reinforcement learning tuned for Solidity and web3 tasks. The live demo generates a cross-chain vault using LayerZero for ETH mainnet and Base L2, produces about 5,000 lines of code in ~40 minutes (core contracts ≈200 lines), compiles, self-corrects to fix a deposit/withdraw bug, and is optimized for Mac/Linux with Windows experimental and CLI/API plans.
AI-written summary · about
chapters (25)
- 0:00Introduction
- 0:28What is Nora? Overview & Form Factor
- 0:55Live Demo: Building a Cross-Chain Vault
- 2:13Error Handling & Self-Correction
- 3:18Reviewing the Generated App
- 4:30Editing and Iterating with Nora
- 5:35Q&A: Model Tuning and Performance
- 6:26Security, Verifiability, and Determinism
- 7:34Prompt Engineering Tips
- 10:07Coding with AI: Best Practices
- 11:24Codebase Indexing & Context
- 12:17Generation Accuracy & Domain Understanding
- 13:18Foundational Models & Post-Training
- 14:45Use Cases & Model Selection
- 15:39Open Source vs Closed Source Models
- 16:49IDE, Traces, and Judging
- 18:08Human Creativity & AI Execution
- 19:44Complex Apps & Human Intervention
- 20:48Chainlink, Oracles, and Integration
- 21:19Platform Support: Mac, Linux, Windows
- 22:15Demo Recap & Usage Quotas
- 24:01Reinforcement Learning & Backend
- 24:39IDE Fork vs Plugin, CLI/API Plans
- 25:31Language & Framework Support
- 26:03Closing & Support Channels
related talks
46mThe amplified developer: zero to shipped in 30 mins with AI Coding Ass
35mVibe Code Your Personal Operating System | AI Productivity with Bethany Crystal
21mBuilding at the Speed of Thought
18mClaude Code & the evolution of agentic coding
45mZero to Shipped in 30 mins with AI Coding Assistants
01mMeet Jòan Varvenne
30mBuild Apps in Minutes 🤯 No Coding Required | Codex Skills Explained
33mNew Renaissance of Building