What Would an Agent-Native Project Workflow Look Like?

Question

Traditional project-management tools are great at organizing work between people.

But AI agents can now do more than provide suggestions—they can investigate issues, write code, run tests, and complete parts of a task.

That changes the workflow.

Instead of simply adding an AI assistant to an existing project tool, an agent-native Linear alternative could treat AI agents as actual participants in the project.

That’s what caught my attention about Sharkly.ai: humans and AI agents can work within the same project workflow rather than keeping agent activity separate from project management.

The bigger question for me is whether project-management platforms should evolve from tracking tasks to coordinating work between humans and AI agents.

Would an agent-native workflow be useful for your team?

Joseph Chung Yin · Answer

The human involvement on AI results must still be reviewed/monitored.  Without it, the entire process may result in lack of accountability when the actions are performed.

Best, Joseph

Mark DE BONT · Answer

Yes, but for now: acting as a governed service for jira board calls. Have a human in the loop to control the results / output. Just to keep a human accountable for every outcome when an agent did the work. Otherwise you get speed with no accountability