Boris Cherny (보리스 체르니), who built Anthropic's AI coding tool Claude Code, said newer AI models have greatly reduced the need to write complex, standardised prompts.

Business Insider reported on Oct. 7 that Cherny recently wrote on X, formerly Twitter, that users should talk to Claude as they would to a co-worker. He said there is no secret to prompting and there is no need to be overly structured or prescriptive for most tasks. He said improvements in model performance have lowered the importance of finding specific phrases or complex prompt formulas.

Cherny presented three key points. He said users only need to tell the AI what it should do, how much effort it should put into the task, and how to verify whether the result came out correctly. That contrasts with past approaches in which users shared various phrasings to draw out a chatbot's performance.

An example prompt Cherny shared was close to natural conversation. After calling a podcast "really amazing", he asked for an artifact based on its contents. He also asked it to use enough tokens and to make the visual materials better than data visualisations from The New Yorker or The New York Times.

A controversial part was the phrase, "You could also try adding watercolours." After one user pointed out that if someone wants watercolours they should give clear instructions, Cherny replied in a way that suggested he trusts Opus 5.5 to decide for itself, saying it has "good taste". The actual output included a beige-and-orange watercolour-style graphic.

Still, it is difficult to view this as meaning that prompt structure is unnecessary for all tasks. Anthropic also recommends structuring inputs and specifying verification methods for complex tasks that handle long documents or multiple materials. Cherny's advice is closer to saying that when using the latest AI, it matters more to clearly communicate goals, effort levels and verification criteria than to rely on detailed prompting formulas.

I am surprised that people are surprised this is how I prompt Claude. Talk to Claude the way you would a coworker. There's no secret to prompting. There's no need to be overly scaffolded or prescriptive for most tasks -- give Claude a goal, and it will figure it out. Back in the Sonnet 3.5 days, your prompt mattered a lot. Nowadays, it's much more important to communicate to the model: 1. What you want it to do 2. How much effort you want it to spend 3. How it should verify that it did the right thing