Insights · Field Notes 02
Agent vs Copilot
In discussions with enterprises about AI ROI, I see a lot of confusion around four terms:
Copilot. Agent. AI-native. Agent-native.
Our definitions are:
- Copilot: a human owns the job; AI assists
- Agent: the agent owns a defined job; the human supervises
- AI-native: software designed for humans to work with AI throughout the workflow
- Agent-native: software designed for humans and agents to operate together
A copilot should be measured by employee productivity. An agent should be measured by the outcome of the job it owns.
And adding AI features to legacy software does not automatically make it AI-native or agent-native.
If your company wants to get measurable value from AI, we need to be precise about what we are deploying and measure the ROI accordingly.
