Most AI-agent workflows still assume one person is interacting with one agent.
But collaborative AI projects often involve several people contributing at different stages.
Imagine two or three teammates joining the same live agent session, adding context, correcting the agent’s direction, or handing the task to someone else without losing the previous work.
For teams building with AI models and agents, this could be useful for things like experiments, debugging, implementation, and research workflows.
I’ve been exploring Sharkly.ai because it takes a collaborative approach to working with AI agents, where multiple people can participate instead of keeping the agent session tied to one user.
The interesting question is:
Should an AI-agent session be treated more like a shared workspace than a private chat?