Both sides
Agent goals and customer goals.
Score what the agent did and what the customer needed, on separate tabs.
← Contact center
Contact center
Coach the call while the customer is still on it.
Playbooks score live calls against your rubric, the QA scorecard fills in on the task, and AI Assist drafts the next reply from your knowledge base. The agent decides what gets sent.
Managers review each score with the transcript beside it, and can change it.
100+ countries · Cloud, hybrid, or onsite · 1 to 1,000 seats
Playbooks
Score the live call against your rubric.
A playbook is your rubric written as goals. Once it's published, it runs beside live conversations, so the agent sees what's left and the manager sees who needs help now.
Categories
Your rubric, in your words.
Group goals by opening and greeting, problem identification, empathy, information gathering, and more.
Scoring
Yes or no, or a scale.
Score each goal yes or no or on a scale, and mark which ones are required.
Publish
Live when you say so.
A playbook doesn't run on live calls until you publish it, and it follows the record type you choose.
QA scorecards
The rubric scores the call. A manager has the last word.
The QA scorecard fills in as the conversation happens and stays on the task, where a manager can check it any time.
- Override. Open the task, read the transcript, and fix a score the rubric got wrong.
- Evidence. The transcript, the recording, and the score sit together on the task.
- Reports. Scores show up in the same reports and on the wallboard, next to handle time and service level.
Bot conversations are scored against the same playbooks as your agents', so a bot that skips a required goal shows up in QA. You fix it by editing its playbook, scripts, or knowledge and publishing the change. How bots work
Live assist
Drafts the reply. The agent decides.
AI Assist works on the live task. It reads what the customer just said and drafts a reply from the conversation, the customer's record, and your knowledge base. That's the difference from a generic chat window: the context is already there, and nobody pastes customer details into another tool.
- Nudge. It nudges the agent toward the next step in the playbook.
- Insert, never auto-send. The agent reviews and inserts. Nothing reaches the customer on its own.
Knowledge base
Answers grounded in the articles your team writes.
Playbooks, vocabulary, and the knowledge base sit together in Setup, and every suggestion traces back to them.
- Write it once.
Your team writes the articles and keeps them up to date.
- From the work you already do.
Knowledge grows from team chat, Meet, and internal calls, with security settings that keep private conversations private.
- One source across channels.
Calls, chats, and emails draw on the same articles, so the answer matches everywhere.
- Words to coach from.
Live transcripts and real-time translation give a coach the customer's actual words.
Questions
What buyers ask about coaching and QA.
Does AI Assist ever send a message on its own?
No. It drafts the reply and suggests the next playbook step. The agent opens the source, reviews the draft, and inserts it.
How does a playbook score a live call?
Each playbook is a set of agent and customer goals, scored yes or no or on a scale, with required goals marked. It tracks every goal while the conversation's happening.
Can a manager change a score?
Yes. The manager reviews it next to the transcript and recording, and overrides it when the rubric got it wrong.
Where do suggested answers come from?
The knowledge base your team writes. The agent can open the source article before inserting the draft.
Publish a playbook and score a live call.
Start a 30-day trial with sample data loaded, publish a playbook, take a call, and watch the scorecard fill in. Then write your own rubric.