Agents as a feature vs agents as the product
In Tulsk, agents have their own runtime (OpenClaw), their own skills, their own scheduling, their own approvals layer, and their own AI project manager (EMA) keeping them honest. The product is built around agents being teammates.
Mention an agent in any task or comment and it reads the context, executes, and replies in-thread.
Linear's Agent platform connects external agents (Cursor, Copilot, Devin, Warp, Factory) into Linear issues. The agent is a contributor while the human stays the primary assignee. The platform is in beta and Agent automations are gated to Business.
If your team already has an agent tool of choice and just wants it visible inside Linear, this works well.