For me the prerequisite for leveraging openclaw was developing function oriented repositories of markdown files tied to my roles that capture pretty much everything I know about the subject and ongoing work, and working with agents as assistants off of those as context. As a founder, product manager, for growth etc.
From there it’s pretty natural that I wanted to talk to an always on agent not tied to any particular machine which has the same context plus access to google drive etc.
One thing that I'm sure of is that the agentic future is test-driven. Tests are basically executable specs the agent can follow and verify against.
When we have solid tests, the agent output is useful and we can trust it. When tests are thin or missing, the agents still ship a lot of code, but we spend way more time debugging and fixing subtle bugs.
Like any CLI Claude Code should follow decades old tradition of providing configurable verbosity levels, like tcpdump's -v to -vvvvv to accommodate varying usage contexts.