Inbox
Where every inquiry lands: how the tabs are divided, what the score means, and what the thumbs actually do.
The Inbox is the raw material. Every conversation the agent has had is here, whole, with the score it earned and why.
The tabs are not the cuts you'd guess
| Tab | What's in it |
|---|---|
| All | Everything except the inactive ones. It isn't literally all |
| Qualified | Anything with a temperature — and that includes Cold. The only ones missing are those with no score yet |
| Pending | The agent is still talking but there's no score yet: not enough data |
| Discarded | Only two reasons put a conversation here: spam and wrong number. A supplier, a journalist or someone asking for a job is not discarded — it gets tagged and stays |
| Inactive | Gone quiet: seven days with no activity, or forty-eight hours if it never became an opportunity. It comes back on its own the moment the customer writes |
The score, read from the list
Each row shows a temperature and a number out of 100. That number measures how complete the data is, not enthusiasm — the full explanation is in Opportunities.
What's useful from here: Pending is not Cold. Cold means "we know things and it's still not going anywhere". Pending means "we don't know anything yet". Treating them as the same thing is how you end up discarding leads that just hadn't answered the second question.
Inside a conversation
The full thread, with photos, documents, and voice notes transcribed to text so you can read instead of listen.
It's read-only. You can't reply from here — that's deliberate, and the badge at the top says so. What you can do is three things:
- Take the opportunity, which assigns it to you, moves it to In contact, and stops the AI agent from replying in that thread. From that point it's yours.
- Rate the agent's replies with 👍 / 👎.
- Hand the conversation back to the agent if a person had taken over.
What the thumbs actually do
Only the agent's messages can be rated — not what a person wrote by hand.
👎 opens a short dialog asking how bad it was and what went wrong. That comment is the valuable part.
Your feedback doesn't retrain your agent by itself. It goes to Trama's team, who read it and use it to calibrate the agent in future versions. A 👎 with a written comment is worth far more than one without: the comment is what says what should have happened, and that's the part nobody can guess from the thumb alone.