You asked an assistant something useful and now a colleague needs to see it. There are two ways to move it: a link to the live conversation, or a copy of the answer itself. They behave quite differently, and the difference matters if the underlying document is protected.
Sharing a conversation
Every assistant answer has a share control in its action row. On a phone it is labeled Share and opens your device's native share sheet; on a desktop it is labeled Copy link and copies the link, confirming with Link copied!.
The link opens a public page showing the full conversation thread — every question and answer up to the point you shared, citations included. The page carries the document's own cover, title and Introduction above the thread, laid out the way the chat page is, so a recipient sees what they are reading rather than a stack of orphaned text. They do not need a Docutrain account.
They can also usually keep going. New questions are added to the shared thread, and the response-style pill and the usual answer actions work as they do in a normal chat. Multi-document conversations keep their multi-document context when shared. And if you shared from a chat opened inside a collection, the link opens that collection with the thread loaded — the recipient lands on the collection page with its cover, document list, and sidebar, and continues there rather than on a separate standalone page.
The share control is hidden when the chat is running as an embedded widget inside another website. Host pages provide their own framing, and a share link out of an embed would land somewhere the embedding site did not intend.
What a share link does not do
A share link does not grant access. This is the part worth being precise about.
The shared page still enforces the document's access rules. If the document requires a passcode, the recipient is asked for it. If it requires sign-in, they must sign in. A document disclaimer, where one is configured, is shown first. Sharing a conversation never exposes a protected document to someone who could not have opened it directly — see access levels explained for how those gates are set.
What the link is, though, is a key. Share links are long random codes and are not listed anywhere public, but anyone holding one can open it, and for a public document that means anyone at all. Treat links to sensitive discussions accordingly. There is one more thing to know: if a conversation is removed by moderation, its share link stops working, and you are not notified when that happens. From the owner's side, shared conversations are not a separate moderation queue — they are the same records that appear in the conversation viewer, so the review and hiding tools there apply to them. See conversations and moderation.
When a shared page cannot be continued
Sometimes the thread appears but the message box is switched off. Docutrain does not leave that unexplained: an amber notice sits above the conversation giving the reason, ending with "The conversation is shown read-only."
| Notice | What happened |
|---|---|
| "The document behind this conversation is no longer available." | The source document has been removed |
| "Some documents in this conversation are no longer accessible." | It was a multi-document conversation and you can reach only part of it |
| "You no longer have access to the document behind this conversation." | The document's access rules changed, or your access was withdrawn |
Two further cases do not turn the page read-only but do interrupt it. If the owner has rotated a passcode-protected document's passcode since the recipient last visited, the shared page discards the stored passcode and asks for the current one, with "Incorrect passcode. Please try again." on a wrong entry — rather than sitting on a blank screen. And a shared thread arrives with its history already filled in, which means it reaches the conversation-length limits sooner than a fresh chat would. The same notices apply: "This conversation is getting long. For a new topic or case, start a New Chat so earlier details are not carried over." while the limit approaches, then the flatly factual "This conversation is long — the earliest messages are no longer being considered." once the earliest messages have actually stopped counting. Each carries a New Chat button that starts a clean conversation with the same document, leaving the shared thread above untouched.
Exporting an answer
If you want the answer rather than the conversation, use the export button — a document icon labeled Export response — in the answer's action row. It opens the Export Response window with two options.
| Option | What it does |
|---|---|
| Copy to Clipboard | Copies the response as plain text |
| Email as PDF | Generates a formatted PDF and sends it to an address you enter |
Choosing Email as PDF asks for an address at the Enter email address: prompt, and Send PDF generates and delivers it, confirming with something like "PDF has been generated and sent to [email protected]!". The PDF contains the formatted response, including tables and, where they were shown, the source citations, labeled with the document's title, category, year, and the owning organization's name. That labeling is what makes an exported answer usable in a report or an email thread later, when nobody remembers which assistant it came from.
The same export window is reused by the Abstract tool's Export as PDF button, so a document's AI-generated abstract can be emailed as a PDF in exactly the same way. Abstracts and the rest of the tool strip are covered in the tools that sit beside the chat.
Note the trade-off between the two routes. A share link is live: it can be continued, and it stays subject to the document's access rules and to moderation. An exported PDF is a snapshot: it survives access changes and document deletion, which is sometimes what you want and sometimes what you should worry about.
Contacting the owner
Sharing an answer often ends in a question the document does not cover, which is what the contact routes are for.
The Contact tool, where the owner has enabled it, appears in the chat's tools area and opens a short form: an introduction line (by default, "Send a message about this document. We'll email the owner with your address as the reply-to."), a Your email field, and a Message field of up to 5,000 characters. Send message delivers it, confirming with Message sent. The owner gets it by email and can reply straight to your address.
Organization landing and gallery pages can carry their own contact window as well, with Your Name, Your Email, Subject, and Message fields. Neither form requires an account.
There is a quiet design decision behind the Contact control that is worth knowing about: it is only shown to readers once the recipient address has confirmed that it is willing to receive that mail. Until then the control simply does not appear. Readers are never handed a form whose messages would go nowhere, and nobody can point a document's contact form at an address they do not control.