/brand-moderator/agent-playbookAgent Playbook
The brand-voice foundation the moderation agent drafts from, tone, canonical response templates, do/don't language, and version history. When a reply gets rejected, the fix goes here, and every future draft improves. This is what keeps replies consistent across thousands of comments and multiple moderators.
What this does for you: The voice foundation every AI-drafted reply is grounded in. Agent Playbook holds your brand tone, approved response templates by category, do/don't language, and the version history of how the voice has evolved. It's the difference between "an AI that replies" and "an AI that replies like your brand."
What's on the screen:
- Brand voice foundations: tone, personality, and the guardrails the agent must respect
- A template library: canonical replies per intent category (complaint, enquiry, appreciation) the agent draws from
- Version history: every change to the playbook, so you can see how the voice was tuned and roll back if needed
- The Reject & Update Playbook loop from the Review Queue lands here: rejecting a draft isn't just a delete, it teaches the agent
Why it matters: Consistency at volume is the whole point. One moderator approving 300 comments a day can't hand-write on-brand replies; the playbook makes every draft start from your voice, and the reject-to-update loop means the agent gets measurably better, the target is ≥70% of AI replies approved with no edit. The version history is also the accountability trail enterprise clients ask for.
When to come here: When onboarding a new brand voice. After a wave of rejected drafts, to codify the fix. When a sub-brand or campaign needs its own tone.
The click-path, step by step
1Playbook controls and editor 
2Playbook content 
- Action HistoryThe auditable log of every moderation action, which reply was approved or rejected, by which moderator, when, and what the AI originally drafted. This is the accountability and compliance trail enterprise and BFSI clients require before they'll let any AI near their social channels.
- AI ResearcherA conversational research agent that runs structured investigations across 30+ external sources on demand. Instead of briefing an agency for a competitive deep-dive or a brand-health read. You pick a template, ask, and get back a deep, sourced answer in minutes, structured as metric cards, channel breakdowns, tables, insights, and citations.
- Review QueueThe single inbox for everything said about your brand, comments, DMs, and reviews across Facebook, Instagram, Twitter, and LinkedIn. Zocket drafts a reply for each one grounded in your Agent Playbook, and a human approves and sends. The non-negotiable guardrail: nothing is ever auto-posted, every reply passes through an approval confirm before it goes live.
- Rules EngineConditional routing logic for the moderation queue. Define what happens to each kind of message, which categories can be auto-drafted, which must be escalated to a ticket, which must never be answered without a human. This is how a client's escalation matrix becomes enforced behaviour instead of a doc nobody reads.