> For the complete documentation index, see [llms.txt](https://docs.filed.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.filed.com/reference/products/binder/chat.md). # Chat ## What Binder Chat is Binder Chat is an AI assistant scoped to the open client's Binder. It answers questions about that client's documents, surfaces what still needs review, and can take actions within the Binder on the reviewer's behalf. Its answers are grounded in the client's actual documents, with citations linking directly to the source. Binder Chat is distinct from the standalone client Chat view (the "Ask" view accessible from the client home screen). Binder Chat is embedded in the Binder's right panel and is aware of the full Binder context: which documents are present, what has been annotated, what is missing, and what the prior year looked like. Binder Chat is available in the web app. ## Opening Chat The Chat panel lives in the right side of the Binder. Clicking the Chat tab in the right panel opens it. The panel includes a header with controls to start a new conversation and to access prior conversation history. **Walkthrough: open and use Binder Chat** 1. Click the **Chat** tab in the Binder's right panel. The empty state shows a greeting and a set of **Suggestions** (starter prompts) such as "What still needs my review?", "Which documents are flagged?", and "What changed since last year?". ![The Binder Chat panel with starter prompt suggestions](https://storage.googleapis.com/filed-prod-public-assets/walkthroughs/binder-chat/image1.webp) 2. Click a suggestion or type your own question in the message box and send it. When the answer draws on specific documents, the response includes citations; clicking a citation opens that subdocument in the center viewer at the relevant page. The message box also supports dictation, so a question can be spoken instead of typed while a document stays open in the center viewer. 3. Use the **+** button in the panel header to start a new conversation, and the **⋯** (more) control to access prior conversation history. Conversations are saved per client and the most recent one resumes automatically when you reopen Chat. ## What you can ask Binder Chat understands the full context of the client's Binder. Starter prompts offered when opening Chat include: * "What's in this binder?" - an overview of the documents present and how they are organized * "What still needs my review?" - documents or workpapers that are open, unflagged, or unsigned * "Which documents are flagged?" - a summary of open flags across the Binder * "Does the income tie to the return?" - an AI-driven reconciliation check against the documents * "What changed since last year?" - new, missing, or changed items compared to the prior year Questions can go beyond these starters. Chat can be asked to summarize a specific document, explain an observation, identify uncategorized items, or locate a figure across the Binder. Before opening individual source documents, it can help to ask a broad question first - such as "what's this client's tax position?" - to get an overview grounded in the client's income sources and estimated position. See [Prompting best practices](/reference/products/chat-and-analyst/prompting-best-practices.md#start-broad-then-go-deep) for more on this approach. ## Cited answers and navigating to sources When Binder Chat answers a question that draws from specific documents, it includes citations in its response. Clicking a citation opens the cited subdocument in the center viewer at the relevant page. This makes it quick to verify any claim Chat makes against the original source without leaving the Binder. ## Getting output in other formats Binder Chat is not limited to written answers. Asking for a table, CSV, or spreadsheet - for example, "give me a CSV of every W-2 and the wages reported on each" - returns the answer as a downloadable file rather than only prose. Use this when the answer is meant to be handed off or used outside of the conversation, not just read. ## Multiple conversations per client A client is not limited to a single ongoing conversation. Starting a new conversation with the **+** button keeps a fresh line of inquiry separate from one already in progress - useful when moving from a general question about the binder to working through a specific finding, without one thread's context bleeding into the other. Every conversation is retained; nothing is lost by starting a new one, and any prior conversation can be reopened from the **⋯** history control. ## Taking actions from Chat Binder Chat is not read-only. It can take actions within the Binder: adding a note or flag, placing a checkmark, signing off a document, or surfacing related marks. When Chat proposes an action, the action is shown as a proposal for the reviewer to approve before it executes. Chat does not modify the Binder without a confirmation step. ## Conversation persistence and history Conversations in Binder Chat are saved per client. Reopening Chat for a client resumes the most recent conversation automatically. Prior conversations are accessible from the conversation history control in the panel header, and a new conversation can be started at any time without losing the history. ## Chat vs. @agent in threads Binder Chat is a real-time conversational interface covering the whole Binder. It is the right tool for broad questions about the client's documents and for taking actions across the Binder. The @agent mention in an annotation thread is a separate mechanism. Tagging @agent in a reply to a note, flag, or other annotation mark triggers a background AI response posted into that specific thread. This is useful when a question is attached to a specific location on a document and the response should live alongside the annotation rather than in a separate chat panel. See [Activity and Sign-off](/reference/products/binder/activity-and-sign-off.md) for the @agent thread flow. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.filed.com/reference/products/binder/chat.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.