How to Design Performance Reviews in an AI-Augmented Workplace

VP of People Ops (VP People Operations)

Our performance review system is broken in the AI era. How do I evaluate someone who uses AI to produce three times the output? Are they three times as good, or are they the same person with a better tool? And what about the person who produces less volume but catches critical errors the AI made? Our entire evaluation framework assumes that individual effort maps directly to individual output, and that assumption just collapsed.

AI Leadership Agent — Performance Strategy

You've just articulated the fundamental challenge that every organization using AI will face — and most haven't realized it yet. Traditional performance reviews are built on an industrial-era assumption: output is proportional to individual effort. AI breaks that equation permanently.

In an AI-augmented workplace, the question isn't 'how much did you produce?' It's 'what did you contribute that AI couldn't?' The answer to that question is what performance reviews should measure.

1. Redesign around three new dimensions: AI leverage (how effectively someone uses AI tools to amplify their impact), judgment quality (the value of decisions that required human expertise), and collaborative multiplier (how much someone improves the work of others, including AI systems). These three dimensions capture value creation better than raw output.
2. Implement 'contribution narratives' alongside metrics. Ask each person to describe three moments where their human judgment made a material difference to an outcome — where they caught something AI missed, redirected a strategy, or applied contextual understanding that no model could replicate. These stories are the evidence of unique value.
3. Evaluate AI skill as a competency, not a cheat code. Using AI effectively is a skill that should be recognized — but recognized as a skill, not as a multiplier on all other competencies. Someone who uses AI brilliantly but lacks domain expertise is a different kind of contributor than someone with deep expertise who uses AI moderately.

When is your next review cycle? How much runway do you have to redesign the framework before it matters?

VP of People Ops (VP People Operations)

Six months. But I also need manager buy-in, and most of our managers are barely comfortable evaluating traditional competencies. Adding AI-specific dimensions feels like it might overwhelm them.

AI Leadership Agent — Performance Strategy

Valid concern. The solution isn't to add complexity — it's to simplify and redirect. Replace some traditional dimensions with the new ones rather than stacking them on top.

1. Reduce total evaluation dimensions. If you currently evaluate ten competencies, cut to six. Remove the ones that AI has made less relevant (e.g., 'attention to detail' when AI handles proofreading) and replace with the new dimensions. Managers shouldn't feel like they're doing more work — they should feel like they're doing different work.
2. Give managers calibration tools. Provide concrete examples and rubrics for each new dimension. 'What does excellent AI leverage look like in a marketing role vs. a finance role?' Don't assume managers can figure this out — give them the framework.
3. Run a pilot with your most adaptable managers first. Let them test the new framework on their teams for one cycle, gather feedback, and refine before rolling out broadly. This gives you proof points and champion managers who can train their peers.

The performance review system you design now will determine whether people see AI as a threat to their evaluation or a tool for their advancement. Get this right and adoption accelerates. Get it wrong and people will hide their AI usage to protect their review scores.

Want to sketch out the six dimensions together? I think we can build something that's actually simpler than what you have today while being more relevant to the AI-augmented reality.