Karpathy's Tip: Rambling to LLMs via Voice is a Great Prompting Strategy
nptacek · x · 2026-07-22
Andrej Karpathy shares a practical pattern for working with LLMs: having a nice long ramble session.
- The Method: When you're too lazy to type out complex context, switch to voice and just ramble for about 10 minutes in total stream of consciousness. LLMs are highly capable of reconstructing these incoherent rambles and often echo your tangled thoughts much cleaner than how you started.
Another developer added that this concept is similar to why auto-injecting context from a RAG vector database is so beneficial. Even if the retrieved information isn't a perfect match, providing the model with extra bits of context significantly helps it grasp your underlying intent.
Related event: Karpathy Recommends 10-Minute Voice Rambling for LLM Context(17 posts)→
More from coding & agent
- Anthropic researcher: 99% of engineers now run swarms of 300+ self-improving agents — AlishaOutridge · 2026-09-11
- Gergely Orosz: Shipping 10x PRs With AI Agents, Sites Fill With Small Regressions — ducha_aiki · 2026-09-11
- Same Echo Maze prompt, three frontier models: all passed visually but shipped the same hidden bug — eyishazyer · 2026-09-11
- Astra storyboards plus Minimax H3 per-shot generation boost video success rates — Hailuo_AI · 2026-09-11
- Codex tip: use Sol with Astra and Luna sub-agents to save usage — pvncher · 2026-09-11
- agents-best-practices: a provider-neutral Agent Skill for designing and auditing agentic harnesses — tom_doerr · 2026-09-11