@_brian_johnson maybe the first step is a more usable UI instead of TUI to surface all these settings and components for users first. A lot of the non-technical users are still scared of all these settings.
Something we kept noticing in my team this year:
most of us were using AI like a smarter Google. Ask, copy, paste.
A small handful had quietly switched — they were getting the agent to do their work. Their work got done in minutes. The rest of us were still typing.
@_brian_johnson maybe someone can soon create a more sounding permission layer to allow non-engineers or non-technical people to build their confidence in using AI agents to actually interact with their files and tools!
@_brian_johnson This is also probably why people are scared of using agents to actually do real work, as they don't know how to control and limit the agents and concerned about the consequences
@_brian_johnson thanks for your feedback! I got some feedback saying that's the key! Will definitely look into how to further improve that part.
Also, agree that we should probably include more details on how to limit the agent on wt it can touch propose also.
Honest feedback most welcome. Contributions even more welcome.
If a chapter doesn't land, tell us — We want to make this better.
https://t.co/l6RgnZuxUT
Every workflow in the book is real — from actual teams. No "summarize this paragraph" toys.
Claude Code is the running example because that's what I use, but every chapter shows the same in Codex, OpenCode, Cursor, Gemini CLI.
Take a look on this 2019-nCov dashboard we just built for the #CoronavirusOutbreak. If you are interested, you can also take a look on our findings at https://t.co/s6gphCQjKg
#coronavirus#DataAnaytics