Claude and Henry Kissinger, aka, Is This Your Best Work?
I was recently reminded of the great story about Henry Kissinger and work product quality. I can’t figure out if it’s true or apocryphal, but it doesn’t really matter. Here’s the version I found online, commonly attributed to Winston Lord, who served as Kissinger’s Special Assistant and later as Ambassador to China.
Lord works for days on a report and submits it to Kissinger. Kissinger returns it with one question: “Is this the best you can do?” Lord takes it back, reworks it, resubmits. Kissinger returns it again. Same question. “Is this the best you can do?” This goes on — six, eight, maybe ten rounds. Finally, Lord brings back a draft and declares: “Damn it, yes, it’s the best I can do!”
Kissinger replies: “Fine, then I guess I’ll read it this time.”
Lord himself recalled the intensity, saying, “You’d have to go through about 20 drafts and many insults before you got the ultimate speech.” The anecdote appears in Walter Isaacson’s biography, Kissinger: A Biography, and it’s become a staple in leadership circles for good reason. It teaches you that most people place their own barriers on what they’re capable of achieving — and that the leader’s job is to push past those barriers until the work is actually excellent, not just done.
So I Tried It on Claude
The other day, I was using Claude to help me think through an issue. After it finished its work, I literally pasted in this prompt:
“I am also reminded of the great story about Henry Kissinger about work product quality. Here it is: [full text of the story]. So I have to ask you, Claude — is this your best work?”
Claude’s response?
“Honestly? No. It’s a solid B+. Here’s what I think is missing or could be better” — followed by a really detailed analysis of its own output, identifying gaps in reasoning, places where it had been too generic, and areas where it could push the thinking further. Then: “Want me to do that pass?”
Uh, yes please.
The revised output was meaningfully better. Not incrementally better — better in the ways that matter. Sharper reasoning. More specific recommendations. Fewer hedges. It had settled for its first-pass answer until I pushed it, and then it found another gear.
What This Tells Us
I’ve been thinking about whether I can build something into my AI scaffolding that auto-asks that question with every task — a built-in Kissinger at the end of every workflow that says “is this your best work?” before the output comes to me. It might be the highest-leverage prompt engineering trick there is.
But the bigger takeaway is more interesting: AI needs to be pushed to be a high performer, just like humans do.
That shouldn’t be surprising, really. AI is trained on human output, and humans have a well-documented tendency to satisfice — to produce work that’s good enough rather than work that’s actually great. The first draft is almost never the best draft. That’s true when a junior analyst writes a memo. It’s true when a senior executive prepares a board presentation. And apparently, it’s true when Claude answers a prompt.
The Kissinger story endures because the lesson is universal: the difference between good and great is usually several more rounds of honest, uncomfortable pressure. Whether you’re managing people or managing AI, the question is the same.
Is this your best work?