> 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/workflow-overview.md).

# Workflow overview

AI Tax Prep runs as a background pipeline. Once started, Filed processes all uploaded documents, builds a data-entry plan, and optionally drives a bot to fill the return in the firm's tax software - all without requiring the preparer to stay on the page.

## Setting up tax prep for a client

The first time a preparer opens AI Tax Prep for a client, an initiation screen appears. It presents two collapsible steps, "Connect a tax software" and "Identify the client in the tax software", with a data-entry toggle below them and a "Run tax prep" button. The screen's subtitle explains the core tradeoff: "Skip any step you don't need. We'll always read the client's forms; connecting a tax software unlocks data entry so we can enter them for you."

**Walkthrough: the tax prep setup screen**

1. From the client hub, open **AI Tax Prep** to reach the setup screen. It has two collapsible steps - **Connect a tax software** (with a Connect tax software button) and **Identify the client in the tax software** - plus an **Enter the return into the tax software** toggle and a **Run tax prep** button. Filed always reads the client's forms; connecting a tax software is only needed to unlock data entry.

![The tax prep setup screen with the first step expanded](https://storage.googleapis.com/filed-prod-public-assets/walkthroughs/tax-prep-initiate/image1.webp)

2. Expanding **Identify the client in the tax software** reveals where you pick the matching client; until a tax software is connected it prompts you to connect one first. You can skip steps you don't need.

![The setup screen with the second step expanded](https://storage.googleapis.com/filed-prod-public-assets/walkthroughs/tax-prep-initiate/image2.webp)

### Step 1 - connect a tax software

This step authorizes the data-entry bot to sign in to the firm's tax software on behalf of the firm. Without a connection, Filed will still read and extract from every uploaded document - the connection is only required for automated entry. Choosing "Skip" moves to Step 2 without connecting.

### Step 2 - identify the client in the tax software

This field is the identifier used in the firm's tax software to locate the client's return. The bot uses it to open the correct return before filling fields. Step 2 is locked until Step 1 is complete or explicitly skipped.

### Step 3 - enable automated data entry

The "Enter the return into the tax software" toggle controls whether the bot runs after extraction. It is grayed out unless a tax software connection is active (or skipped) and a client ID has been entered. When the toggle is off - or when Steps 1 and 2 were skipped - only extraction and planning run.

## Running tax prep

**Walkthrough: run tax prep**

Click **Run tax prep**. Filed redirects to the active pipeline view, which shows "Running tax prep" with a live processing indicator (for example, "Extracting tax forms"). The page updates automatically as each phase completes, so you can leave and come back.

![The active tax prep pipeline view with a processing indicator](https://storage.googleapis.com/filed-prod-public-assets/walkthroughs/tax-prep-running/image1.webp)

## How Filed processes the documents

After the pipeline starts, a real-time progress view appears showing each phase as it completes. The pipeline works through these stages:

**Document processing and extraction.** Filed reads every uploaded document, splits multi-form PDFs into individual sub-documents, and classifies each by form type and issuer or owner. Extraction runs in parallel across all detected forms.

**Reconciliation.** The AI compares extracted data against the prior-year tax software backup when one is available. It builds a per-field data-entry plan, resolving sources of truth and flagging conflicts it cannot resolve automatically.

**Planning and finalization.** Filed finalizes the forms list, generates Pre-Entry Notes (issues, missing documents, excluded documents), and writes the summary. The pipeline view shows each phase with a spinner while active and a check mark when complete.

Processing time varies by document volume. The preparer can navigate away at any point; the pipeline continues in the background and the view updates automatically when each phase completes.

## Reviewing results

After processing completes, four areas of the product are populated:

* **Summary** - current status and, after data entry, entry results by field. See [Summary Tab](/reference/products/ai-tax-prep/summary-tab.md).
* **Forms** - all tax forms the AI detected, organized by category. See [Forms Tab](/reference/products/ai-tax-prep/forms-tab.md).
* **Pre-Entry Notes** - issues, missing documents, and excluded documents. See [Pre-Entry Notes Tab](/reference/products/ai-tax-prep/pre-entry-notes-tab.md).
* **Source Files** - every uploaded source document with a viewer. See [Source Files Tab](/reference/products/ai-tax-prep/source-files-tab.md).

## Automated data entry

If the data-entry toggle was enabled at setup and the pipeline completed planning successfully, the data-entry bot runs after extraction. The progress view shifts to a data-entry display showing the current phase:

* **Setup** - the bot is preparing and building the data payload.
* **Entering** - the bot is actively filling forms in the connected tax software.
* **Verifying** - post-entry validation is running; the bot checks its own entries and retries fields where the written value does not match the expected value.

The view updates automatically. The Backups panel is visible throughout, showing the current backup count.

### Task status indicators

The client tile and the AI Tax Prep view both reflect the pipeline's current state:

| Status               | What it means                                                              |
| -------------------- | -------------------------------------------------------------------------- |
| Processing           | The pipeline is actively running.                                          |
| Ready for data entry | Extraction and planning are complete; the data-entry bot has not run yet.  |
| Done                 | The full process is complete, including data entry if it was enabled.      |
| Failed               | The pipeline encountered an error it could not recover from automatically. |

When the pipeline fails during data entry, the view shows the specific error and the recommended recovery action. The error codes and their meanings are:

| Error                 | What it means                                                               | Recovery                                                                                        |
| --------------------- | --------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------- |
| INVALID\_CREDENTIALS  | Filed could not sign in to the tax software.                                | Update the software connection before retrying - repeated failed sign-ins can lock the account. |
| CAPTCHA\_FAILED       | The login check could not be passed automatically.                          | Usually clears on its own; try again in a few minutes.                                          |
| CLIENT\_NOT\_FOUND    | The client ID entered at setup does not match a return in the tax software. | Check the client ID against the software, then try again.                                       |
| CLIENT\_ALREADY\_OPEN | The return is open in another session.                                      | Close it in the other session, then try again.                                                  |
| NOT\_REGISTERED       | The tax software is not set up on the workstation.                          | Register the software on the workstation, then start tax prep again.                            |
| TIMEOUT               | The tax software did not respond within the expected window.                | Usually clears on its own; try again in a moment.                                               |
| UNKNOWN               | The pipeline paused for an unexpected reason. Nothing is lost.              | Start again - Filed will pick up where it left off.                                             |

## What happens next

After AI Tax Prep completes, the recommended next steps are:

* Review the [Summary Tab](/reference/products/ai-tax-prep/summary-tab.md) to see field-level data-entry results.
* If Flagged fields appear, open the Binder to inspect each flagged entry against its source document.
* If Manual fields appear, enter those items directly in tax software - they come from form types the bot does not currently support.
* When a draft return PDF is available in the Summary, click "Run AI Review" to have the AI check the completed return for issues.


---

# 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/workflow-overview.md?ask=<question>&goal=<endgoal>
```

`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.