How Terry learns
What Terry takes from your edits, your results and your outcomes, and what it never changes on its own.
Terry gets better from the work your team already does. Emails that earn replies shape its voice. Documents that appear in replied emails count for more. Your edits and your outcomes are kept as examples. One thing never changes: Terry does not rewrite your rules, knowledge or agent on its own. Anything it suggests waits for a person.
Your best emails become its examples
When Terry writes for an agent, it looks at that agent's past emails that went out and got the best results, the ones that earned a reply or a meeting first. It matches their voice, length and shape without copying them.
Emails that drew a no, an unsubscribe or a complaint are left out, and so are people marked lost or not a fit. So the more your team sends good emails and records outcomes, the better Terry's examples get.
Editing and rejecting drafts
- When you edit a draft before you approve it, Terry records how much you changed and keeps your final words as an example of what good looks like.
- When you approve a draft as it is, that counts as a draft that needed no changes.
- When you reject a draft, Terry records that too.
- When a draft is held, the reason is recorded.
These judgements are kept as test cases for how Terry writes. From them, Terry's nightly check may propose changes, such as a new rule or a knowledge addition. They appear as Suggestions at the top of Settings, Knowledge. Press Review to open them in the knowledge console. Nothing is applied until an admin approves it there. A rule change goes live when approved; a knowledge change then waits for approval under Rules & knowledge.
Knowledge that works rises
When Terry picks which parts of your knowledge to use, documents that appeared in emails that got replies rank higher. Documents that appeared in bounced or rejected emails rank lower.
Record outcomes
Outcomes tell Terry and your team what happened after the reply. Open the conversation in the Inbox, press Move, then Record an outcome, and press one of the buttons:
| Button | What it records |
|---|---|
| Meeting held | The meeting happened. |
| No-show | They did not turn up. |
| Lost | The deal is lost. Terry stops any flow for this person. |
| Not a fit | They are not a fit. Terry stops any flow for this person. |
For Lost and Not a fit, Terry asks Why? One line helps the next rep. Home shows the week's outcomes.
See what is working
Analytics is its own item in the menu (Pro plan and above). The card What is working gives the clearest finding from the last 90 days of sent email. Press All splits, or choose View: All detail, to see every split: by flow, by list or source, by time of day, by weekday and by subject length. Each group gets a verdict: working, not working, too early or level, with the numbers in plain sentences. A short version sits on Home. These are counts, not AI guesses.
Terry's read on a conversation
In the Inbox, open a conversation that has a reply and press More (the three dots), then Ask the coach. Terry's read tells you Why this matters, the Next move, and a Recommended reply you can use. Human approval is still required: you send it.
Campaign reviews
When Terry's flow reviews is on (an admin switches it under Settings, Features), Terry reviews each live flow every two days and writes what worked and what did not, with the numbers. You decide what to change.
What Terry does not do
- It does not edit your rules, knowledge or agent settings by itself.
- It does not send because it learned something. Every send still passes your approval settings and the usual checks.
Related
In Classic
- Suggestions are under Suggestions in the knowledge console, from Settings, Knowledge, Manage in ContextBus.
- Outcomes: open the conversation in the Inbox and use the outcome buttons beside it.
- What is actually working is a card on Analytics.
- Terry's read sits beside an open conversation in the Inbox.
- Campaign reviews: Terry's campaign reviews on the Agents page, under What Terry does on its own.