Every thread, pre-read for you

A workspace-side AI fluent in your product, your customers, and your numbers. Ask for the metric, the recap, or the reply — the groundwork's already done.

One assistant, every chore

It lives in the inbox and shows up already briefed — with the metric, the queue state, the language, and the reply.

Plain-language questions

Type 'which channel spiked yesterday?' and the live numbers come back from your own workspace.

Finds the missing article

It combs conversations for questions your docs never answer — ranked by frequency, drafts included.

The recap is waiting

A 50-message thread opens with the brief already written — and it updates itself as messages arrive.

No language border

Inbound lands in your language, outbound leaves in theirs; neither side notices the seam.

The reply, pre-written

Drafts assembled from your own sources — review, adjust, send.

Shared brain, shared meter

Copilot drafts and your customer-facing agent read the same sources and draw from the same AI-request pool — no per-operator surcharge, ever.

Answers before the switch flips

Today's capabilities, the billing model, and the early-access roadmap.

Automatic three-bullet briefs on every conversation, inbound translation into the operator's language with a one-click "Show original" toggle, and outbound translation of drafts into the customer's. Knowledge-grounded reply drafting and free-form questions are in early access — a demo gets you in.
Each new customer message triggers a background rewrite covering the last 50 messages — what is wanted, what has been offered, where things stand — as three bullets of fifteen words or fewer. Briefs sit above the thread and in the case details, in each operator's own language, and cover bot-run threads and Zendesk-synced tickets alike.
Operators pick from fifteen interface languages for briefs and translation. On the customer side, detection is automatic and translation runs in both directions. Up to fifty messages translate per pass, and caching means reopened threads render instantly.
That is the heart of early-access drafting: retrieval runs over the sources linked to your agent, and every draft cites what it used, so an operator can verify before sending. The same sources power the customer-facing agent — one library, two uses.
No. Its actions draw one-for-one from the same AI-request quota as customer-facing replies — no per-seat add-on, no hidden meter, one pool in your usage dashboard. And since teammates are unlimited, every operator on the team gets it.
Yes — Copilot attaches to the conversation, not the channel. Briefs generate for widget chats, email, and Zendesk sync alike, and translation works on any thread in the inbox.

Never open a thread cold again

Recaps and translation work today — start the trial and read your first brief.