Intercom closed conversations

Each closed conversation. When it closed, reopens and first human reply. Inbox, channel, whether Fin resolved it, and the rating's score. Never messages, contacts or rating comments.

Access

Kind: Sign in on the vendor's page; nothing typed here
What to give: Access token
Why that one: An Intercom access token that can read conversations. This adds no cost. It counts what your AI spend produced.

What it attributes on its own

From the source: conversations closed per day, at the moment each last closed; reopens, counted apart so a conversation closed twice is still one conversation; when each conversation opened, to give the time to resolve; the inbox it sat in, by name; whether a person ever replied, so a deflected conversation is visible; who resolved it: Fin alone, Fin then a person, or a person, where Intercom sends it; its topic: your Topic or Category attribute, else Intercom's topics; how it came in, as Intercom names it; the rating, 1 to 5, never the remark
Needs a rule or a trace: which conversations an assistant other than Fin handled: Intercom does not record it. The cost per assisted conversation is at least the cost per closed one; the workflow, which you tell us; how hard a conversation was: a one-line answer and a week-long thread both count as one

The addresses it calls

https://api.intercom.io/conversations/search
https://api.intercom.io/teams
https://api.eu.intercom.io/conversations/search
https://api.eu.intercom.io/teams
https://api.au.intercom.io/conversations/search
https://api.au.intercom.io/teams
https://api.intercom.io/auth/eagle/token

Valistry calls only these addresses. It blocks any other call and logs every call.

Connecting it

Create an Intercom app. Developer Hub → New app. Then click Create app. Open ↗
Copy the access token. valistry-read → Configure → Authentication. Copy the Access token. Open ↗

If your assistant writes the conversation id into a trace, ask us, and we will match each conversation to its exact cost instead of by workflow.