Ask three more times: Anthropic co-founder Ben Mann on getting the most out of Claude

Key points

Key takeaways from the Lenny's Podcast interview with Anthropic co-founder Ben Mann (August 2026):

Use new tools as new tools, not old ones. Mann's core advice for getting ahead of AI is to be ambitious in how you use the tools and willing to keep learning them — people who treat new tools as if they were the old ones tend not to succeed.

Be ambitious with the ask. The difference between people who use Claude Code effectively and those who don't is whether they request the ambitious change rather than a timid one.

If it fails, ask three more times. When a request doesn't work the first time, retrying — or completely starting over — has a much higher success rate than banging on the same failed attempt again and again.

No one is immune to job replacement. Mann notes that even he, at the centre of the transformation, is not immune — at some point AI is "coming for all of us."

Anthropic itself runs on Claude. Mann has said the vast majority of Anthropic's own code is written by Claude, letting a much smaller team be dramatically more impactful.

Safety-first by design. Mann co-founded Anthropic after helping architect GPT-3 at OpenAI, driven by the mission to build aligned, helpful, harmless and honest AI — the approach he credits with shaping Claude's personality.

Read more: Lenny's Newsletter