Don’t Aim to Replace Yourself with AI
From Glean founder Arvind Jain’s 20VC interview: “replace yourself with AI” is the wrong goal. When everyone has the same tools, winners raise the bar—and strong companies may get larger.
A popular workplace slogan says everyone should use AI to replace themselves—spend freely on tokens, automate your job, and move on. Jain’s reply on 20VC is blunt: that is the wrong target. It overrates what today’s AI can fully own, and it misunderstands how competition works when rivals have the same tools.
Ninety percent is not a win
Pick almost any role, he suggests, and ask whether AI can replace it end to end. An executive assistant is a fair example: AI can help book flights and shuffle calendars. It still struggles with the invisible judgment that makes someone trusted—tone, priorities, when to push back, what not to put in writing.
If your competitor also has AI, and they keep a sharp human on top of it, a “90% AI assistant” is not a moat. It is table stakes. The edge is the remaining judgment—plus what you build with the time AI frees up.
Shrink to hold still—or grow to leap
Jain pushes back on the idea that AI’s main corporate story is a smaller headcount doing the same work. If you and your rival both have the same AI stack, and you cut people to freeze today’s output, they can do the same. If instead they hire to chase a product ten times better, they may outgrow and outrun you.
Productivity per person should rise. The competitive bar rises with it. In that frame, the goal is not “how few humans can we keep,” but “can we place talent on the projects that create outsized value.” Asked about Glean’s own future size, he hopes—if things go well—for a much larger company in five years, not a hollowed-out one.
Composite roles, thinner pure-execution jobs
What does change, in his view, is how work is packaged. One person may increasingly cover slices of engineering, product, and design to ship something end to end. Sales may blur too: negotiation, demo, and solution design in fewer hands. Heavy specialization becomes harder to justify when AI handles more glue work.
Roles that are mostly pure execution—taking a ticket, pulling a number, refreshing a dashboard—are more exposed. Business leaders will increasingly ask systems directly. Recruiting “sourcers” who only hunt résumés may fold into fuller-cycle hiring roles. The point is not panic about every job title; it is noticing which work was always thin on judgment.
A better personal target
So if “replace yourself” is the wrong slogan, what should replace it? Something closer to Jain’s competitive logic: use AI to remove busywork, keep ownership of judgment, and aim the freed capacity at work that is ten times more valuable than yesterday’s baseline. The model is a lever. The strategy is still human.