Boris Cherny says concise prompts unlock more in Claude Code
Technology
Boris Cherny, the mind behind Anthropic's Claude Code, has a simple tip: don't drown AI chatbots in super-detailed instructions.
At a Y Combinator event, he explained that giving too much direction can actually hold back what these advanced models can do.
Instead, just set clear goals and basic rules. No need for an essay.
Cherny: models keep revealing new abilities
Cherny pointed out that AI is moving fast: what worked six months ago might already be old news.
He backed up the idea of "lazy prompting" (thanks to Google Brain co-founder Andrew Ng), which means keeping instructions short and focused on what matters most.
Interestingly, Cherny shared that his team keeps finding new abilities in their models, so there's still a lot more under the hood than we realize.