You can now edit each AI helper prompt individually through personas, limit access to specific groups, set different LLMs, etc.