Chat Interface

Starting a Conversation

There are several ways to begin a new chat:

• Home Page — Type a message in the input box or click an assistant card.
• Sidebar — Click the New Chat button.
• Agents Page — Browse all assistants and click one to start a conversation.
• Existing Conversation — Continue any previous conversation from the sidebar or chats page.

When you select an assistant, you’ll see its welcome page explaining what that assistant can help with. What else appears depends on how you got there:

• Starting a new chat (from the home page or the New Chat button) shows just the assistant’s introduction and the message input — a clean slate.
• Opening the assistant from the Agents page also lists your previous conversations with that assistant, so you can resume one instead of starting over.

Required Context

Some assistants are configured to require specific context before they will answer — for example, your company size, industry, or strategic focus. When this is the case, an amber banner on the welcome page lists what you need to provide. Send your first message including that information. If anything is missing, the assistant replies with a short clarification asking for the missing pieces; once you provide them, the conversation continues normally and the requirement is not re-checked for the rest of that conversation. To review or update the context you provided, click the clipboard icon next to the message input. The panel shows your original context along with any updates you’ve added, and lets you submit a new amendment from the same place.

Sending Messages

Type your message in the input area at the bottom of the chat and press Enter to send (or Shift+Enter for a new line).

Attaching Files

Click the paperclip icon to open the attach menu:

• Upload files — opens your computer’s file dialog. After you pick files, a window shows a short notice about sensitive documents and the files you picked (add more there with Browse or by dropping them, or take some off the list). Nothing is uploaded until you press Attach to chat.
• Attach from parsed documents — pick a document that has already been read instead of uploading it again (see Attaching a document that is already parsed).
• Parse in the background — a switch for this conversation, on by default. While it is on, every file you attach goes to the background parser, whatever its size (a plain-text .txt or .md file is the exception: it is always read in place). It takes a few minutes to start, and the chat stays responsive. Turn it off and an ordinary small file (up to 1 MB, other than a spreadsheet) is read right away on the server instead (a file over 1 MB, any spreadsheet, a scanned PDF and any file the server cannot read itself still go to the background parser). Each conversation keeps its own switch, remembered in your browser (a different browser or device starts with it on). The one you set before your first message carries into the conversation that message creates, and the next new chat starts with the switch on again.

Files you attach:

• Maximum file size: 25 MB per file
• You can attach multiple files to a single message
• The background parser takes at most 10 of your files at once (an administrator can change this): files waiting for their turn, being read, or waiting for your Read with OCR or Skip choice all count, across all your conversations and the copilot. A file you attach past the limit is refused: its chip says Failed with the reason (You already have the most files the parser takes at once. Wait for some to finish, then try again.), and a message you send does not carry it. Remove that chip and attach the file again once some of your files have finished. A plain-text file (.txt, .md) is never refused for this reason, and neither is an ordinary small file while Parse in the background is off (the server reads it right away).
• Remove files before sending by clicking the X on each attachment (a file that is still being read stops being read)
• Supported formats include PDFs, Word documents, text files, and more

Each attachment appears as a chip inside the message box, above where you type. It shows a badge for where the file came from (Upload, Background, Parsed or Ingested), its name, and where it stands: Files you have attached but not sent stay with the conversation while you move around the app, so you can switch to another conversation and come back. Reloading the page clears an unsent draft, as it clears the text you typed. When the parser finishes a file you attached in the chat, a notice also appears (“Agreement.pdf” is ready, or “Agreement.pdf” could not be read), even if you have moved to another conversation by then. A long read, such as a scan read by OCR, can finish while you work elsewhere. Sending while a file is being read. While a file is uploading, waiting for the parser or being parsed, Send is held. With Parse in the background on, which is the default, that is every file you attach except plain text, so a message with a file can wait for the parser to start (turn the switch off to have ordinary small files, spreadsheets apart, read right away). The box says Waiting for 1 file to be read…, with a Send without it link that sends your message now and leaves that file attached for your next one. A file that is ready goes with the message. So does a file that could not be read: the assistant is told, by name, that it could not be read, instead of answering as if it had. A scan waiting for your choice, or being read with OCR, never holds Send: it stays attached for your next message, and the box says so (1 file is not part of this message. It stays attached for your next one.).

Scanned PDFs

The pages of a scanned PDF are pictures, so there is little or no text to read until OCR (optical character recognition) reads them.

• A scan of up to 2 pages is read by OCR straight away.
• A longer scan waits for you: its chip says Scanned PDF (12 pages) and asks Read with OCR (~2 min) or Skip. You can keep chatting meanwhile.
• Read with OCR queues the scan, then shows the time it takes. When it is done the chip turns Ready. Send your next message and the assistant has the text.
• Skip leaves the scan attached, but unread. Its chip then says Scan not read or Failed with scanned PDF, not read, depending on whether the text the scan carried of its own (a signature line, a stamp) was kept. Either way the assistant is told that it is a scan nobody read, or that it could not be read, and does not answer from it. To use the document, attach a text version (a Word file, or a PDF whose text can be selected), or attach the scan again and choose Read with OCR.
• A question you leave unanswered is skipped for you after 24 hours.

Attaching a document that is already parsed

Attach from parsed documents opens a picker with up to three lists:

• My uploads — files you uploaded in the chat, in the knowledge-model copilot or in Test Ingestion that the parser has already read. A plain-text file (.txt, .md) attached in the chat is read in place and is not listed.
• Parse only — files from data sources set to Parse only: they were read and their text was saved, but they were not added to the knowledge graph.
• Ingested — files from data sources that were ingested into the knowledge graph and whose document is attached to an entity there.

Each tab says this in one line under the tabs, and again when you hover it. A file whose artifact type has no entity that attaches the document is ingested but has no document in the graph, so it is in neither of the two data-source lists (see Async Document Parsing). Everyone has My uploads. Parse only and Ingested are for people who can read data sources (the Data Source Managers role), and an ingested document is listed only when your own access to the knowledge graph lets you see it. Search by name, filter by file type (on Parse only and Ingested also by data source, job and, on Ingested, artifact type), tick one or more documents and press Attach. Each filter takes one or more values: a document is listed when it has any one of a filter’s values and matches every other filter and the search. A picked document is not read again, so its chip is Ready at once. Each row shows the page count, the size, when the document was parsed and where it came from (Your upload, or the name of the data source and its job, for example Job 113). An eye icon on the row opens the document’s text before you attach it (see Previewing the text of a file); it does not tick the row. A document that cannot be attached is not listed. The picker lists only the documents it can attach. A Parse only document whose stored text is no longer there is simply missing from the list: it is not greyed out, and nothing counts it. An Ingested one is missing in the same way only while graph access control is enforcing; otherwise it is listed. A list with nothing to attach says No documents available to attach., except Ingested, which says Nothing here yet. A document is listed once it has been ingested, its artifact type attaches the document to an entity, and you may read it in the graph. While graph access control is enforcing, the stored text of the document must also still exist. An administrator can check the Jobs page and Artifact Types. A file under My uploads carries its own text, so it is listed for as long as the file is kept. To bring a missing document back, re-ingest it or upload the file again. Cannot find a document? Go through the checklist I can’t find my document in the picker. While a list loads, and when it cannot. Changing the list, the search, a filter, the file type or the page clears the rows at once and shows a loading state until the new rows arrive; the rows of the list you left are never shown under another. If a list cannot be read, the picker says so with Try again and shows no count. The Ingested list needs the knowledge graph to check which documents you may read; while the graph cannot be reached, that list says We couldn’t check which documents you may read right now. Try again in a moment. My uploads and Parse only do not need the graph. A count that ends in a plus (2,000+ documents) means the list covers only the newest documents: search, or choose a data source, job, artifact type or file type, to reach an older one. When an attach fails. Attach tries every ticked document and does not stop at the first one that fails. The documents that went in are attached and leave the selection. If some did not, the dialog stays open and one error box, These documents could not be attached:, lists each of them with its reason. They stay ticked, so you can press Attach again, or untick one to take its line off the list. The list is a first check: a document it lists can still be refused, for example when your access to its source ended or its stored text went after the list was drawn. A picked document stays under the access rules of where it came from, for every person who reads the conversation: someone who cannot read that data source, or that document in the graph, gets nothing from it, and you lose it too if your own access ends. What the assistant reads says how that is checked. A document from My uploads is your own file, available to the conversation’s members once you have sent it.

What the assistant reads

The assistant works from the text extracted from an attachment. How much of it the assistant gets depends on how you attached the file. A file you upload is put into your message as text, and only the first 10,000 characters of that text are. On a long document that means the assistant is reading the opening section, not the whole file — the remainder is not passed to it. When an attachment is cut short this way the assistant is told the document is partial and asked to say so rather than summarizing it as though it were complete. A document you attach from parsed documents (from My uploads, Parse only or Ingested) is not put into your message as text. The message carries only a reference to it, and the assistant reads the document through its document tools: it indexes the document first, then searches it or summarizes it. The index covers the first 100,000 characters of the document, so a search finds a fact deep in a long document. A document longer than that is searched only up to that point. Because the assistant reads a picked document this way, your access to its source is checked each time it reads the document, not only when you attach it. When the person the assistant is answering may not read the source, such as a member of the conversation without access to it or a person whose access has been removed, the assistant gets not available instead of the text. That holds in a later message and in a fork of the conversation too. After you send, the attachment stays on your message — click it to open it in the panel on the right. A file you uploaded opens as it was uploaded, and you can download it from there if you need a copy; a document you picked from parsed documents opens its text. The small eye next to it opens the text the file was read as, in the same panel. Your original file is unaffected by the 10,000-character cut of an upload: it stays on the message in full and you can still open and download it. Files attached before this was introduced still show on the message, but their originals can’t be re-opened: only their extracted text was kept at the time, so clicking them, like their eye, opens that text (see Previewing the text of a file).

Previewing the text of a file

The eye icon (tooltip Preview parsed text) opens the text a file was read as, so you can check what the assistant works from. You find it:

• on a chip in the message box that says Ready or Scan not read, beside its X. A chip that is still being read, waits for your OCR choice, failed or was cancelled has none;
• on each row of the Attach from parsed documents picker that can be attached;
• beside every file attached to a message in the conversation, once it is sent. Clicking the file’s name opens the file in the panel on the right (the original, or the text for a file that has no original); the eye always opens the parsed text.

The text opens in the panel on the right when you click the eye on a chip, in the message box or on a sent message: the same panel the file itself opens in, with the file name, a Parsed text label, and the buttons to expand and close it. From a row of the picker it opens in a window over the picker instead, because the picker covers the page. The panel and the window show the same thing:

• Loading. It opens on the first 20,000 characters. Load more adds 100,000 each time, and Showing 20,000 of 85,598 characters says how far you are.
• Header. The file name, then the number of characters, the number of pages when it is known, the parser, and OCR used when the text came from OCR. Scan not read and Truncated (the file was cut while it was read, so this is only part of its text) are shown in amber.
• Pages. Where the parser marked the pages, a Page 3 divider stands where the marker was.
• Search. Search the loaded text highlights the matches (the first 500 at most), whatever the capitals, and counts them (2 of 14, or 1 of 500+ when there are more than 500). The arrows, or Enter and Shift+Enter in the box, move between matches. It searches only the text loaded so far: when nothing matches and more text is waiting, it says Not found in the text loaded so far. Load more to keep searching.
• Copy. Puts the text loaded so far on the clipboard, with a Page 3 heading where each divider is.
• Nothing is kept. The text is held only while the panel or window is open, and the browser is told not to store it. Closing it, opening another file in the panel or moving to another conversation drops the text, and opening it again starts from the first 20,000 characters.

Either way it shows the whole text the parser stored. That can be far more than the assistant works from: it is given the first 10,000 characters of an upload, and indexes the first 100,000 characters of a picked document (see What the assistant reads). A plain-text file (.txt, .md) is the exception. It is read in place, and only its first 10,000 characters were ever kept, followed by a notice that the rest was cut. The preview of a longer one ends there, and its header says Truncated. A scan that was not read. In the main chat, the eye on a Scan not read chip answers This scan was not read (no OCR was run). and shows no text. In the picker and in the knowledge-model copilot, the window opens on the little text the first pass found (a signature line, a stamp), with Scan not read in amber in its header. Who can open it. Before you send, you. After you send, whoever may read the file on the message: the people with access to the conversation, and staff accounts. A picked document follows the rules that decide what the assistant reads (see What the assistant reads): its source is checked each time anyone opens it, so a member without access to it, or you once your own access has ended, gets the message below instead of the text. When there is no text to show, the panel or window says why instead of staying blank:

• You can no longer open this document (it was removed, or you lost access to its source). The file is gone, or your access to the document’s source or to the conversation has ended. For a picked document it can also mean that the access check could not run just then; try again later.
• This file is still being read. Its text can be opened once that is done. or This scanned file is waiting for a choice about reading it with OCR. Retry asks again once the file is ready.
• The reason the file could not be read, when it failed.
• The parsed text of this file is no longer stored. The parsed copy of a picked document has been deleted.
• This document’s text is no longer stored; re-ingest it or upload the file. From the picker’s eye: the stored text of a data-source document went after the list was drawn.
• Could not check your access right now. Try again. From the picker’s eye: the knowledge graph could not be reached to check your access to an ingested document. Retry asks again.

Selecting a Model

If multiple AI models are available, use the model selector next to the input area to choose which model processes your message. Your selection persists for the current assistant session.

Generating Documents

If document templates are configured, a Generate button (wand icon) appears next to the model selector. Click it to browse templates organized by category and generate presentations, formatted messages, and more. See Document Templates for full details.

If the Connection Drops

If the chat loses its connection to Experio (a network drop, a server restart or a deployment), it reconnects by itself, for as long as the outage lasts. It waits a little longer between tries, up to about 30 seconds, and tries again at once when your browser comes back online or you return to the tab. You do not need to reload the page. While it is reconnecting:

• Your conversation, the sidebar and what you typed stay where they are. A refresh that fails behind them does not replace them with an error screen.
• Sending is refused, with the message Reconnecting… Your message was not sent. Nothing is lost: your text, the files you attached and any flow you picked stay in the message box, an edit of your last message stays open, and a Required Context form keeps its answers. Send it again once the chat is back.
• An answer that was being written is stopped. Ask again once the chat is back.

When the connection returns, the conversation and the sidebar refresh themselves within a few seconds.

Reading Responses

AI responses stream in real-time — you’ll see the text appear as it’s generated. When you ask a follow-up in a long conversation, your question moves to the top of the view and stays there while the assistant works, so you can keep reading it alongside the progress steps. The view follows the answer once it starts arriving.

Task Progress

For complex queries, the AI may perform multiple steps (searching documents, analyzing data, synthesizing information). A task progress indicator shows:

• The current step being performed
• Completed steps with checkmarks
• Sub-steps and tool calls being made

The indicator updates as the assistant moves through the run, so a long answer shows where it has actually got to rather than a single unchanging label. A typical question moves through Analyzing Question, Routing Query (deciding how to handle it), Planning, Scope Analysis (finding the data that answers it), then Report Generation as the answer is written. Steps vary by question — a query that is answered from what it already found will skip the retrieval step entirely.

Citations and Sources

When the AI references your organization’s documents, you’ll see numbered citations in the response. Click any citation to open the Source Details Panel on the right, which shows:

• File title and type
• Data source name (Box, Google Drive, SharePoint)
• File path and last modified date
• A content preview of the referenced section
• A direct link to the original file

Citations are filtered to your own access, like the answers themselves — you will only see a citation for a record you are allowed to read. On very broad questions a correct answer can arrive with no citations at all; see Sources and citations. For graph entities behind an answer, open the node in Explore Graph from the Provenance or Graph Context tab to see how properties were extracted, imported, or enriched. See Graph Lineage.

Answer tabs

An answer has up to four tabs above the response: Answer, Steps, Provenance, and Graph Context. Steps shows what the assistant did to write the answer. Provenance and Graph Context appear for answers that draw on the knowledge graph. The steps the assistant took also appear in the reasoning section above the answer while it works.

Artifacts

When the AI generates long-form content (reports, analysis, structured outputs), it opens automatically in the Artifact Panel on the right side. You can:

• Maximize the panel for full-width reading
• Minimize it back to a side panel
• Resize it by dragging the handle between the chat and the panel; the panel remembers the width you give it on this browser and opens at that width the next time
• Close it to return to the chat view
• Export the content as PDF, DOCX, or Markdown — click the download icon in the panel header.
• Generate document from the content — click the wand icon in the panel header and pick a template to have the agent produce a formatted file (DOCX, PPTX, or PDF). See Document Templates.
• Edit the content inline with the built-in editor

You can also export from the message action toolbar on the chat response itself (see Export & Sharing). Generated files (DOCX, PPTX, PDF, CSV, and similar) use a Download button in the file panel to save the original file.

Questions from the Assistant

Sometimes the assistant needs something from you before it can answer — which part of the business you’re asking about, whether you want a summary or a full report, which of several close matches you actually meant. Rather than writing the question into its reply and leaving you to guess the format of an answer, it asks with a question card underneath the response. A card carries a handful of questions — five at most by default, and an administrator can set a different limit per assistant — each with a few suggested answers shown as selectable chips:

• Pick one or several. Most questions accept more than one answer. A question whose options are mutually exclusive — summary versus detailed report, say — accepts only one, and picking a second replaces the first. The hint under each question tells you which it is.
• Add your own. Type in the Add your own… box and press Enter (or click +) to add an answer that isn’t offered. Your additions appear as chips you can remove with their X. If what you type matches one of the options, that option is ticked instead of adding a duplicate.
• Send answers submits the whole card in one go. It stays disabled until every question has at least one answer.

What you send is posted as a normal message in the conversation — one line per question, in Question: answer form — so the exchange stays readable when you scroll back. The assistant then resumes from wherever the question came from: usually it goes back and retrieves data using what you told it, and occasionally — when the question was only about how to present an answer it already has — it simply rewrites the answer. You don’t have to use the card. The message box stays open, so you can type your answer instead and send it like any other message; the assistant reads it as the reply to what it asked. Either way, the card clears as soon as you send. If you reload the page or come back to the conversation later, a question you haven’t answered yet is still waiting — the card is restored with the conversation. It disappears if you stop or delete that turn. The assistant won’t keep asking indefinitely. After a couple of rounds of questions — two by default, configurable per assistant — it answers with what it has and explains in plain language what it could not pin down.

Recommendations

After each response, the AI may suggest follow-up questions based on the conversation context. Click any suggestion to send it as your next message. Suggestions are held back while a question card is waiting for you — nothing appears under an open question but the card itself. They return once the assistant delivers its final answer. Suggestions, filter options and the title of a new conversation are prepared after the answer is complete. If they are not ready within 90 seconds (an administrator can change this), they are skipped, and a new conversation can keep the title New Chat.

When the Knowledge Graph Cannot Be Queried

If the knowledge graph is unreachable, too busy, or a read takes longer than its time limit, the answer says so plainly: that the knowledge graph could not be queried this time, and that you can ask again in a moment. It does not report this as no data or nothing found, because when a lookup fails nobody knows whether matching records exist. If part of the data was retrieved before the failure, the answer is written from that part and says it may be incomplete. The error text itself is never shown. An administrator can change the time limit with GRAPH_READ_TIMEOUT_MS (see Graph Read Time Limit).

When an Answer Takes Too Long

An answer that runs past its time limit (270 seconds unless an administrator changed it) is stopped, and the conversation shows This answer took longer than the time limit, so I stopped it. Please ask again in a moment. The message is saved in the conversation like any other answer. Answers started with the Deep Research toggle are not limited this way.

Data Updates in Progress

When an administrator has turned the data-update notice on, a note reading New data is being added right now, so answers may take longer than usual. appears under the progress indicator while your answer is written, and stays under the latest answer after it is finished, until you send your next message. The step that carries it is called Data update in progress. It shows when files are being ingested or enriched at the moment you ask. It is only a heads-up; the answer is written as usual. The notice is off by default (see Chat Turn Limits and the Ingestion Notice).

What the AI can see

Answers are limited to what you are allowed to see. When your organization has access control enabled for the knowledge graph, every question the assistant asks of the graph is filtered to your own access before any answer is written — so two people asking the same question can correctly get different answers. You do not configure this and there is nothing to switch on. If an answer seems to be missing something you expected, it usually means that record is not linked to you, and an administrator can check with Preview as user on the Access Control screen.

Agent Flows

Some requests are handled by an agent flow — a multi-step pipeline the assistant runs for you (for example, drafting a response to an RFP). There are two ways to start one:

• Just ask. When your request matches a flow, the assistant confirms first (“I can run RFP Response Pipeline — proceed?”). Once you approve, it launches the flow in the background, so you can keep chatting while it works.
• Pick it from the composer. Click the Launch a flow button (workflow icon) next to the message input to see the flows available for this conversation, each with a short description. Choosing one launches it right away — the pick is your go-ahead, so there’s no separate confirmation. The composer prefills a short prompt you can extend with any details the flow needs before sending. The flow you picked shows as a pill on the top edge of the message box, so it’s clear the next message will launch it; clear it with the pill’s X.

Files you’ve attached to the conversation are routed into the flow automatically — including the file you attach to the message that starts it, so “here’s the RFP, draft a response” works in one go. Only files you uploaded are picked up this way; a document an earlier run produced is never swept in as the next run’s input unless you ask for it by name. If the flow still needs a file you haven’t shared, it doesn’t start, and a message in the conversation names what to attach — by the input’s readable name (for example RFP document), not an internal field name — and tells you to attach it and send your message again. A file you attached that is still being read has no text yet, so the flow does not start on it: the assistant tells you to wait a moment and ask again. If the file is a scan waiting for your choice, choose Read with OCR (or Skip) on its chip in the message box first, then ask again. If the file could not be read, attach it again. Only you can choose OCR: the assistant never starts it for you. Asking the assistant to continue a run that already stopped doesn’t do what it sounds like, and it now says so. No run can be picked up where it left off once it has ended — that includes one that finished successfully, not just one that failed or was cancelled. Running the flow again starts a new run from the first step, with no partial output carried over, no reading or revising of the earlier run’s document, and every approval asked again. Before it launches anything, the assistant looks up what actually happened to your last run in this conversation, and tells you plainly that this is a fresh start rather than describing it as a continuation. If that earlier run is still going, it reports where the run has got to instead of starting a second one.

Watching a flow run

When a flow launches, a flow-run view opens automatically in the panel on the right and follows the run live. It is deliberately spare — it exists to tell you the flow is working and to let you act when it needs you:

• A status badge showing where the run is — Pending just after you start it, Running while it works, and Awaiting review whenever there’s a question or an approval waiting for you. It can read Paused for a moment while a pause is still making its way to the panel
• When the flow is reading your documents in the background (a scan is read by OCR), the badge reads Waiting for the parser (2 of 5 files) instead, and the line below it says the flow carries on by itself once your documents are read. Nothing is asked of you, and you can keep chatting
• One line naming what it is working on right now, by the step’s name: Working on Draft the response
• The approval or question card when the run pauses for you, so you can answer it without leaving the conversation (see Approval Steps). Documents and files a card shows you open right there in the panel
• A Stop button to cancel a run that’s still going

When the run finishes, the panel closes — unless you’re reading a file the run produced, in which case it waits until you close that document rather than taking it away mid-read. Nothing is lost either way: the result and any files the flow produced arrive as a message in the conversation, and so do the questions you answered and the decisions you made — the conversation is the record. You can close the panel at any time — the flow keeps running, and a View progress control appears above the message box to reopen the live view. While a flow is active, its conversation pulses in the sidebar, so you can tell at a glance that work is still in progress; the pulse clears once the run finishes. If a run is waiting on you (see Approval Steps), the reopen control turns amber — “Review needed — open to approve or reject” — and the conversation’s sidebar row shows an amber review badge instead of the pulse, so you can spot which conversation needs action from anywhere. These sidebar signals keep up live, with no page refresh: the pulse appears as soon as a flow starts, gives way to the amber badge when a run pauses for you, returns to the pulse once you’ve answered and the flow carries on, and clears when the run finishes or you Stop it. This works both for the conversation you’re reading and for conversations working in the background. If your connection drops while a flow is running, the live view and these signals reconnect by themselves and catch up, so you do not need to refresh. You can also ask the assistant for the status or results of a flow it launched at any time. When the flow finishes, a brief notification appears and its full result — along with any files it produced — is added as a message in the conversation on its own, without a manual refresh. Files stay with the answer that produced them: each one is listed inside its own message, so running a flow twice in the same conversation gives you two answers with their own files rather than everything piling onto the newest one. Re-running a flow doesn’t replace the earlier document either — when a run would write a file that another run already produced here, it keeps its own copy under a slightly different name. Filenames mentioned in the answer text are clickable too, and open the same reader. A run can also stop because the service running it did — a deployment or a restart. A run stranded that way used to keep its Running status indefinitely, leaving the conversation looking permanently busy over work nothing was doing. Those runs are now reconciled when the service starts up: a run with nothing left to finish it is marked as failed, the sidebar pulse clears, and the conversation is free again. A run parked at a review gate is never touched — a question waiting on you survives a restart, along with the amber badge that tells you it’s there. Nor is a run waiting for the parser: it carries on once its documents are read. Passing trouble no longer sinks a whole run either, though the cover has edges worth knowing. A momentary network problem in a step is retried once — never in the steps where an AI agent is doing the work itself, since re-running one of those could repeat something it had already done, and otherwise only while the run is working on its own. Once it has stopped to ask you something and picked up again from your answer, nothing in that stretch is retried, because re-running a step there could put a question you have already answered back in front of you. So in a flow that asks for your sign-off early, the work that comes after your sign-off is not covered by that retry. A step that failed and was then genuinely redone no longer forces the whole run to be reported as failed either — as long as it really was redone: the same step, over the same piece of work, on a later pass. A different step covering similar ground later on doesn’t clear it, and neither does anything else; a failure that nothing redid still fails the run.

Asking about a document a flow produced

You can carry straight on in the same conversation — “summarise the response you just wrote”, “what did it say about pricing?” — without re-attaching anything. The content of documents the flow produced is read into the conversation alongside the messages, so the assistant can answer from them directly. What it gets read in that way is a bounded excerpt, not the full document. There are limits on how much of any single document, and how much across the conversation as a whole, is carried this way; the most recently produced document is served first, so the one you just asked about is the one that fits. Where a document was cut short, or an older one didn’t fit at all, the assistant is told — so it can say the answer is based on part of the document rather than answering confidently from a fragment. That excerpt is no longer all it has to go on. The assistant can also list the documents flows have produced in this conversation, search inside one, and read a section of it at a time — so a question about something deep in a long report is answerable even though only the opening was read in. Searching looks for an exact word or phrase and ignores capitalization; it is not a semantic search, so name a term you expect to actually appear in the text. Each document is checked against your own access before it is read, and only documents produced by flow runs in this conversation can be reached this way. You can still open any of them in the panel and read them yourself.

Approval Steps

Some flows pause partway through for a person to sign off — for example, before a draft is finalized. These pauses don’t interrupt your chat, and you don’t have to go anywhere else to clear them: the question comes to the conversation. When a run pauses, a notification appears and the flow-run panel shows an Awaiting review card where you can approve or reject it right there, with a comment (optional when approving, required when rejecting). The question lands in the conversation too, as its own message — one per gate, naming the gate it belongs to and carrying a short excerpt of what’s up for review; for a mid-run question, the question itself and any options it offers. A run that pauses on three gates at once posts three messages, so you can tell them apart. Long documents scroll inside the card, so the approve and reject buttons stay in reach. A flow can also ask you a question mid-run — type your answer into the card and send it to continue. Acting on the card resumes the flow and the card clears. You can change what you were shown before approving it. Where the card presents a document or a block of text, an Edit button turns it into an editable field — reword a section, or edit a document’s title and each of its sections — and then Approve and continue. Tables and lists appear as you see them, and you click into a cell or a line to change it; formatting the editor can’t keep opens as text instead. A part you don’t change goes on exactly as it was, and a part you edit is saved as standard Markdown. The version you edited is the one that carries on to the rest of the flow, not the one you were originally shown. An edited card is marked edited while you work, and Revert puts the original back. If a run pauses on more than one question at once, you don’t have to wait between them. Answer them in whatever order you like: an answer you give while the flow is still busy with a previous one is held and submitted automatically as soon as the run is free, so you don’t have to sit and retry. An answer can still be refused: if that question is no longer waiting for one — someone has already answered it in the Agent Inbox, or the run has moved on — a short message tells you so rather than leaving you wondering. On the rare occasion the run stays busy long enough that the automatic retries give up, the card comes back so you can answer it again. Answered cards clear as you go and don’t come back — though a flow that revises its work can put the same gate up for review again with the new version, which arrives as a fresh card and a fresh message rather than the answered one returning; the ones you haven’t answered stay on screen and stay usable. Each card is signed off on its own, carrying whatever you edited on it — there is no bulk sign-off, since an approval and a rejection each need their own note. Until the last question is answered, the conversation keeps its amber review badge in the sidebar. Once you’ve acted, your decision is recorded in the conversation — which gate it was, whether you approved or rejected it, and any note you left. It is folded into the question’s own message, under Your response, instead of arriving as a second message further down: scrolling back, you read the question and the answer it got as one entry, and the question drops its “answer it in the run panel” instruction, which your answer has already satisfied. If you edited before approving, the record says so and includes what you approved, so reading back through the conversation later shows the version that actually went through rather than the draft you were offered. A rejection with no reason given is recorded as such. Answers to a mid-run question are merged the same way, under the question they answered. A long response — an edited draft you approved, say — is shown as a preview with a View full response button that opens the whole thing in the panel on the right. There is one deliberate exception. If you sent a chat message of your own in between — asking something else while the run was parked — your answer stays as a separate message below it rather than folding back up into the question. Nothing you answered later ever appears above something you said earlier. The same pending review is also available to admins in the Agent Inbox, but you don’t need it — either path resumes the flow, which then posts its outcome back into the conversation.

Message Actions

Hover over any message to access its action toolbar.

AI response toolbar

Open More (⋯) on an AI response for additional actions:

User message toolbar

Stopping a Response

If the AI is generating a response you no longer need, click the Stop button that appears during streaming. The response will stop at wherever it has reached.

Editing Messages

Click the edit icon on your last message to modify it. The AI will regenerate its response based on your updated message. This is useful when you want to refine your question without starting a new conversation.

Conversation Management

Renaming

Click the conversation title in the header to edit it. The new name saves automatically when you click away or press Enter. Press Escape to cancel.

Forking

To copy the whole conversation, open the three-dot menu in the header (or on the row in the sidebar, chats list, or a folder) and select Fork. To branch from a specific AI reply, use Fork from here on that message. See Forking Conversations.

Sharing

Click the Share button in the conversation header to share with other users:

1. Search for users by name or email
2. Assign a role: Viewer (read-only) or Editor (can contribute)
3. Manage existing shares — change roles or remove access

Deleting

Use the conversation menu (three dots) to delete a conversation. A confirmation dialog will appear before deletion.

Filters

Some assistants support response filters that let you narrow results by category or type. When available, filter options appear above the message input. Click a filter to apply it to your next message. Like follow-up suggestions, filters are hidden while the assistant is waiting on a question card.