AI Control Centre Settings
The AI Control Centre is where your team configures how the AI behaves inside the Task Manager — what it does automatically, how aggressively it acts, and how it handles different types of customer interactions. Here's what each setting does:
Classification description — Instructions that tell the AI how to recognise and categorise incoming communications. Used by the Linker when deciding which category to assign a new task to.
Task Playbook — The operational instructions the AI follows when planning actions on a task. The AI reads this every time it suggests next steps for tasks in this category.
When a new communication comes in, the AI runs a short sequence automatically: It first decides whether the communication belongs to an existing task or needs a new one. Then, if enabled, it reads the context of that task and proposes what should happen next — suggesting actions, pre-filling draft replies, and updating task details like status or due date.
Two layers of control
Company-wide switch — A master on/off switch for AI planning. When turned off, the AI still classifies and links communications to tasks, but stops suggesting actions or adjusting task details.
Per-category settings — Each task type (Dispute, Promise to Pay, etc.) has its own AI configuration. You can turn the AI off entirely for a category, set it to only adjust task details without suggesting actions, or run it in full mode where it does both.
Note on system categories: The ten built-in categories can be edited but not deleted. Once you edit a system category, it's marked as customised — meaning future platform updates won't overwrite your changes.
Custom rules — Additional constraints layered on top of the playbook. Up to 20 rules per category.
Allowed actions — The action types the AI is permitted to suggest for this category. If an action type isn't on this list, the AI cannot suggest it regardless of what the playbook says.
Reset to defaults — Restores the category's classification description, playbook, and allowed actions back to the original out-of-box configuration. Only available for the built-in system categories.
AI Playground
The Playground is a safe testing environment. It lets you test how the AI would handle a specific scenario before it goes anywhere near your live task list.
You paste in a sample email or call and see exactly what the AI would do — which task it would create or link to, what actions it would suggest, and what draft it would write — without any of it actually happening.
The Playground shows you:
Which task category the AI would assign (and why)
Whether it would create a new task or link to an existing one
Which actions it would propose
What draft content it would generate
A plain-English explanation of the AI's reasoning at each step
