> For the complete documentation index, see [llms.txt](https://help.filed.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://help.filed.com/reference/products/ai-tax-prep/pre-entry-notes-notifications.md). # Pre-entry notes notifications Every item in the Observations list carries a severity level: Critical, High, Medium, or Low. The level tells the preparer how urgent the issue is and what kind of response it typically requires. ## The severity levels Filed uses four severity levels, from most to least urgent: Critical, High, Medium, and Low. * Critical and High generally indicate a conflict or data problem that, if left unaddressed, could make the return incorrect. These warrant a decision before or during tax software entry. * Medium and Low generally indicate an ambiguity or potential issue the AI is surfacing for awareness. The AI is being transparent about something it could not resolve definitively, and these may not require action. ## Higher-severity notifications (Critical and High) A higher-severity notification represents a detected conflict or discrepancy that needs a human decision before the return can be considered accurate. A common example is an SSN mismatch: the SSN on the 1040 does not match the SSN on a Form 1098. Filed cannot determine which is correct automatically. Opening the source documents side by side lets the preparer find the correct value and confirm it in tax software during or after entry. Other situations that typically generate higher-severity notifications: * Conflicting values across two documents for the same field (EIN, name, dollar amount). * Extracted data the AI has low confidence in due to document quality or ambiguous formatting. * Fields that require a human decision before a value can be written to the return. ### How to respond to a higher-severity notification **Walkthrough: respond to a conflict observation** 1. In the binder's **Observations** view, find the observation that flags a conflict - for example, "W-2 Box 3 (Social Security wages, $102,000.00) differs from Box 1 (wages, $102,200.00) by $200.00." Each card shows the source document and the AI's full explanation; use **Show more** to read the complete reasoning. 2. Click **View in source doc**. The source document opens with the observation pinned as an annotation on the exact field, and the AI's note appears in the Notes & flags panel on the right. Compare the figures on the document to decide the correct value before it is entered. ![An observation opened in its source document with the AI note pinned to the field](https://storage.googleapis.com/filed-prod-public-assets/walkthroughs/observations/image2.webp) Resolve higher-severity items before the data-entry bot runs where you can. An unresolved conflict means the bot may write an incorrect value or skip the field. ## Lower-severity notifications (Medium and Low) A lower-severity notification reflects something the AI noticed but could not definitively classify as a problem. The preparer's judgment is required. A common example is a potential K-1 duplication: the AI sees two K-1s from what appears to be the same partnership. Upon review, they are from two separate partnership interests with different EINs and separate schedule breakdowns. Both are legitimate and two screens are needed, so this is not a real duplicate. Other situations that typically generate lower-severity notifications: * Apparent duplicates that are actually distinct items, for example two K-1s from related but separate entities. * Amounts that look unusual compared to prior year but are legitimate current-year figures. * Edge cases the AI is surfacing for human confirmation rather than acting on autonomously. ### How to respond to a lower-severity notification **Walkthrough: review and resolve a low-priority observation** 1. Open an informational observation - for example, "W-2 Box 5 (Medicare wages, $102,000.00) same as Box 3 - consistent. Informational." Read the AI's reasoning, and use **View in source doc** if you want to confirm the figure against the original. 2. Decide the outcome. If the item is a false positive, click **Dismiss** to remove it from the list. To leave a note for the file or a colleague instead, click **Comment** and reply in the thread that opens in the Notes & flags panel. ![The comment thread on an observation with a reply composer](https://storage.googleapis.com/filed-prod-public-assets/walkthroughs/observations/image3.webp) If a lower-severity item turns out to be a real issue after review, correct it. If it is a false positive, note it for the file and proceed without action. ## Finding notifications in the UI **Walkthrough: open the Observations list** 1. From the client hub, the Binder tile shows an "N Observations" badge. Open the Binder and select **Observations** in the left panel (it also carries a count). 2. The **Document observations** view opens - a consolidated list of every finding the AI surfaced while sorting the documents, grouped by document category (Income, Payments, and so on). Each card shows the source document, the finding text, and the actions **View in source doc**, **Dismiss**, and **Comment**. Every observation is also pinned as an annotation on the page it came from. ![The Document observations list grouped by category with per-observation actions](https://storage.googleapis.com/filed-prod-public-assets/walkthroughs/observations/image1.webp) The consolidated list groups observations by document category rather than by a severity column; open an individual card to read the AI's full explanation and decide how to respond. ## Binder cross-reference After data entry completes, the same issues that generated these notifications are also visible as colored annotation marks in the Binder, anchored to the exact page and location in the source document where the AI detected the discrepancy. The Binder's Observations screen shows a consolidated list. --- # 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://help.filed.com/reference/products/ai-tax-prep/pre-entry-notes-notifications.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.