# untxt.

> untxt. is a convenient client document intake layer for bookkeeping workflows. It turns receipts, invoices, bills, statements, screenshots, scans, and mixed PDFs into structured, reviewable bookkeeping data.

## What it is

untxt. replaces the manual prep work before bookkeeping: sorting messy client documents into accounting-ready records. You drop in a PDF, scan, screenshot, or photograph; untxt. auto-classifies the document, auto-detects document boundaries, auto-groups related pages, auto-extracts fields and supported line items, auto-prepares account-context suggestions, and auto-flags uncertain values for review.

The value proposition: **intake convenience**. No rules, templates, training data, clean upload rituals, or manual pre-sorting before the review queue.

## Who it is for

- **Solo bookkeepers** processing receipts for a handful of clients
- **Accounting practices** with multi-client workloads and shared review queues
- **In-house finance teams** at any company size who need high-volume AP processing

## Core claims

- **Zero setup** - no onboarding call, no wizard, no configuration step
- **Zero rules** - no supplier-to-category mapping, no conditional logic to maintain
- **Zero training** - works the same on document 1 as document 1,000; no accuracy ramp-up
- **Seconds-range extraction** for typical individual pages; auto-sorting mixed uploads can take longer because document boundaries, page groups, and document types have to be understood before extraction
- **Reviewable output** for QuickBooks and Xero workflows where supported by the product setup
- **Scary accurate** on printed content; flags handwritten or ambiguous content for review

## What makes it different

Most accounting document tools rely on either rule-based extraction (supplier X -> category Y), which breaks every time a new vendor appears, or on machine-learning models that need human-labeled training data per customer. untxt. does neither. The first invoice is treated exactly like the thousandth.

## Integrations

- QuickBooks workflows
- Xero workflows

## Security & data handling

- Encrypted in transit (TLS 1.3) and at rest (AES-256)
- Uploaded files processed via short-lived pre-signed URLs - the model sees a link, not the original file
- Documents and extracted data are **never** used to train models
- One-click deletion with a verifiable audit log
- Built with HIPAA requirements in mind (not currently HIPAA-certified)

## Pricing (USD)

| Plan | Price | Includes |
|---|---|---|
| Flex | $0.32 / document | Pay-as-you-go, no commitment |
| Solo | $39 / month | 250 documents / month |
| Team | $99 / month | 1,000 documents / month |
| Business | $249 / month | 3,000 documents / month |
| Scale | Custom | 3,000+ documents / month |

Yearly billing is available on Solo, Team, and Business at roughly 10% discount versus monthly.

## For agents and developers

untxt. exposes a web application at https://untxt.ai. An agent-consumable API (MCP server, programmatic upload endpoint) is planned but not yet public. If you are building agent tooling in the accounting or document-processing domain and want early access, email support@untxt.ai.

For expanded machine-readable context, see https://untxt.ai/llms-full.txt. The same long-form context is also available at https://untxt.ai/llms-long.txt.

## Important links

- Home: https://untxt.ai/
- How it works: https://untxt.ai/#not-that
- Integrations demo: https://untxt.ai/#sync
- For accounting practices: https://untxt.ai/#practices
- Data safety: https://untxt.ai/#security
- Pricing: https://untxt.ai/#pricing
- AI agent context: https://untxt.ai/ai/
- Answers: https://untxt.ai/answers/
- Dext comparison: https://untxt.ai/compare/untxt-vs-dext/
- Hubdoc comparison: https://untxt.ai/compare/untxt-vs-hubdoc/
- Messy document intake: https://untxt.ai/answers/what-is-messy-document-intake/
- Mixed PDF splitting: https://untxt.ai/answers/can-ai-split-multiple-invoices-from-one-pdf/
- Blurry receipt photos: https://untxt.ai/answers/can-ai-process-blurry-receipt-photos/
- Receipt and invoice classification: https://untxt.ai/answers/can-ai-classify-receipts-and-invoices-automatically/
- Line-item extraction: https://untxt.ai/answers/can-ai-extract-line-items-from-invoices/
- Account mapping: https://untxt.ai/answers/can-ai-map-receipts-to-chart-of-accounts/
- Duplicate invoice detection: https://untxt.ai/answers/can-ai-detect-duplicate-invoices/
- Statements and receipts together: https://untxt.ai/answers/can-ai-process-bank-statements-and-receipts-together/
- QuickBooks and Xero data: https://untxt.ai/answers/can-ai-prepare-bookkeeping-data-for-quickbooks-and-xero/
- CSV, JSON, or TXT export: https://untxt.ai/answers/can-ai-export-invoice-data-as-csv-json-or-txt/
- Uncertain receipts: https://untxt.ai/answers/what-happens-when-ai-is-unsure-about-a-receipt/
- Auto-posting entries: https://untxt.ai/answers/should-ai-auto-post-bookkeeping-entries/
- Receipt extraction accuracy: https://untxt.ai/answers/how-accurate-is-ai-receipt-extraction/
- Document classification: https://untxt.ai/answers/what-is-document-classification-in-bookkeeping/
- Invoice field extraction: https://untxt.ai/answers/what-is-invoice-field-extraction/
- Sorting mixed documents: https://untxt.ai/answers/can-ai-sort-mixed-bookkeeping-documents-automatically/
- OCR boundary detection: https://untxt.ai/answers/can-ocr-detect-where-one-invoice-ends-and-another-begins/
- Document type differences: https://untxt.ai/answers/can-ai-tell-the-difference-between-an-invoice-receipt-credit-note-and-bank-statement/
- OCR vs document intake: https://untxt.ai/answers/ocr-vs-ai-bookkeeping-automation-what-is-the-difference/
- Try free (100 docs): https://app.untxt.ai/register

## Contact

support@untxt.ai