In my mind Co-pilots, Agents, and AI Teammates are different.
I look at co-pilots as representing assisted intelligence: They can provide suggestions to help users complete tasks but, ultimately, humans control them and make final decisions. A good example is GitHub Copilot which makes quick suggestions for software developers to incorporate into the programs they’re writing.
Agents, on the other hand, represent autonomous intelligence. They have decision-making capabilities, but are still very task-oriented and require very little ongoing interaction with users. A good example is one of our portfolio companies, Lyft. They have an AI-powered agent which can autonomously handle refunds for customers without involving any human representatives.
Finally, there are AI teammates, taking us into a new era of collaborative intelligence. The idea here is that these teammates collaborate with humans, then adapt and learn in order for them to achieve shared objectives. One example could be an AI marketing teammate which collaborates with human users by doing research and providing suggestions. It may be able to autonomously take some actions, while still leaving some tasks and decisions to be made by the human user.
