
Jay Sen Lon
September 4, 2026

Somewhere in your firm's document pile is a supplier invoice none of your tools can read. Or a bank statement long enough to eat half a credit pack.
That is where Dext, AutoEntry, and Hubdoc stop looking alike. Their pricing pages suggest three versions of the same product. Their behavior on a 30-line wholesale invoice says otherwise. Before you renew or switch, here is which one extracts line items, which one charges extra for them, and which one never will.
TLDR:

Dext is a document capture tool for accounting firms and small businesses. It digitizes receipts, invoices, and bank statements before they land in the ledger, then codes them with supplier rules: you set how a supplier's documents should be coded once, and Dext applies that rule going forward. It publishes to Xero, QuickBooks, Sage, and 30-plus other accounting platforms.
Pricing runs on document allowances tied to user bundles. The Business plan is around $25 a month billed annually, or roughly $31.50 month to month, for five users and 250 documents. Practice plans are priced by client count and start around $240 a month for 10 business clients. Extraction is strongest on standard, printed, English-language documents.

AutoEntry is owned by Sage and built mainly for the UK, Ireland, and other Sage-heavy markets. Pricing runs on credits, from Bronze at $14 a month for 50 credits up to Sapphire at $469 for 2,500.
Credits are spent per document type, not per upload. A standard invoice costs 1 credit. The same invoice with line items extracted costs 2. A bank statement costs 3 credits per page. Unused credits roll over for 90 days, then expire. Firms with detailed invoices or long statements burn through an allowance faster than the plan name suggests.
AutoEntry publishes to Sage, Xero, QuickBooks, and FreeAgent. Firms comparing AutoEntry alternatives usually name the credit model as the reason they started looking.

Hubdoc is Xero's document capture tool. Xero bought it in 2018 and bundled it free with most Xero subscriptions. Non-Xero users can run it standalone for about $12 a month, and it also publishes to QuickBooks Online and Bill.com. Its signature feature is fetching: connect a bank or a utility and Hubdoc pulls the statements and bills in itself, from over 700 institutions.
Extraction stays at the header: supplier, date, total, invoice number. No line items. Coding automates through Supplier Rules, set once per supplier, per client. Because it costs nothing on top of Xero, many firms run Hubdoc for years without ever comparing Hubdoc alternatives.
Hubdoc gives you the total and stops. Every line still gets typed by hand once the document lands in the ledger.
Dext will give you the lines, but only on its higher plans, and they draw against the plan's document allowance. AutoEntry will give you the lines for 2 credits instead of 1.
For a firm processing a 30-line wholesale invoice with mixed tax rates, this matters more than the sticker price. Two of the three tools can extract the lines. Both make you pay for the depth, one way or another. Line-item depth becomes a line-item cost.
| Tool | Pricing structure | What it costs you in practice |
|---|---|---|
| Dext | Business plan around $25/month billed annually for 5 users and 250 documents; Practice plans from around $240/month for 10 clients | Cost scales with users and clients, not document volume; line items only on higher plans |
| AutoEntry | Monthly credits, from $14 for 50 to $469 for 2,500 | 1 credit per invoice, 2 with line items, 3 per bank statement page; credits expire after 90 days |
| Hubdoc | Free with a Xero subscription; about $12/month standalone | Header-only capture at every price point; the cost is bookkeeper time on line items |
Three different meters. Dext bills you for growing your client list, even when documents per client stay flat. AutoEntry bills you for document complexity, so the same invoice costs double the moment you want the lines. Hubdoc bills you nothing and hands the work to your team instead. The Tofu vs Dext breakdown shows how these structures play out over a year.
Dext extracts from English and the major European languages, and works best on printed Latin-script documents. Hubdoc's stated coverage is English only.
AutoEntry's own document guidelines list Arabic, Chinese, and Japanese as unsupported. Documents in those scripts fail to process rather than extract with errors. All three accept standard PDF, JPG, and PNG uploads, but the script on the page decides whether anything comes out.
If your clients buy from suppliers in Bangkok, Shenzhen, or Dubai, the question is not price. It is whether the tool reads the document at all. None of these three was built for that, which is why firms in that position end up searching for multi-language receipt OCR software.
Dext meters bank statement extraction separately from the document allowance. The Business plan includes 10 statement pages a month, and pages beyond that are billed per page. AutoEntry charges 3 credits per page, so a 12-page statement costs 36 credits before you have touched an invoice. Hubdoc fetches statements automatically and stores them, but it does not extract the transaction lines. For Xero clients the bank feed covers that. For everyone else, the statement gets keyed by hand.
Batch files are the other snag. Dext and AutoEntry can split a PDF page by page, but none of the three works out where one document ends and the next begins. A scanned batch of receipts with a two-page invoice in the middle still needs sorting by hand before upload.
Dext and AutoEntry both automate coding through supplier rule building: you process a document from a given supplier, correct the coding once, and the tool applies that same treatment to future submissions from that supplier. This rule sits at the supplier level, so a new vendor requires a fresh rule before its documents code automatically.
HubDoc follows a similar structure through Supplier Rules, configured manually per client. Each client's suppliers need their own rule set built before HubDoc applies coding without manual work, and rules configured for one client's supplier do not carry over to another client using the same supplier.
Across all three tools, automated coding depends on a rule existing first. Until a supplier has been seen and a rule built, documents from that supplier need manual coding on every submission. A side-by-side Tofu vs Dext comparison shows how rule-based and AI-learning approaches differ in setup overhead., and firms onboarding a new client typically rebuild the same rule set from scratch even if the client's supplier base overlaps with an existing client.
Tofu, an AI bookkeeping automation platform for accounting firms, starts from the other end. Every line item on every invoice, by default, on every plan. No toggle, no extra credit, no higher tier. A 30-line wholesale invoice with mixed tax rates gets coded in full. "What used to take me 3-4 hours can be done in 30-60 minutes," says Tammy Tan at Klozer.
Language coverage runs to 200+ languages, including handwriting, so the Thai, Chinese, and Arabic documents AutoEntry lists as unsupported get the same extraction as an English PDF. Bank statements process at any length with no per-page charge. Multi-document PDFs split automatically, so the batch of scanned receipts goes in as one file.
Pricing is flat: $79 a month for Pro and $199 a month for Business, both with unlimited users. No per-user multiplier and no credit pool. See Tofu pricing for what each plan includes.
Setup starts with Tofu reading your chart of accounts and coding history, then learning from each correction. There is no rule-building step before automation kicks in, and a new client does not mean starting over.
| Capability | Dext | AutoEntry | Hubdoc | Tofu |
|---|---|---|---|---|
| Line items | Higher plans only | 2 credits per invoice | Not available | Default, every plan |
| Bank statements | 10 pages included, then per page | 3 credits per page | Fetched and stored, no line extraction | Any length, no per-page charge |
| Languages | English and major European | English; Arabic, Chinese, Japanese unsupported | English | 200+, including handwriting |
| Batch PDFs | Page-by-page split | Page-by-page split | Manual | Automatic boundary detection |
| Coding setup | Supplier rules per client | Supplier rules per client | Supplier rules per client | Learns from coding history |
| Pricing model | Per user bundle and per client | Monthly credits | Free with Xero | Flat $79 or $199/month, unlimited users |
Dext fits a firm already on one of its 30-plus integrations, processing mostly English-language printed documents, and comfortable with a bill that grows as the client list does.
AutoEntry fits a UK or Ireland practice already on Sage, as long as someone tracks credit spend against document mix. A credit shortfall shows up mid-month, right when the bank statements land.
Hubdoc fits a Xero-only firm whose supplier invoices are simple enough that supplier, date, and total are all you need to publish. It ships with the subscription and asks nothing else of you.
None of the three fits a firm with suppliers in Bangkok, Shenzhen, or Dubai, or one that codes 30-line wholesale invoices every week. For that firm the constraint is not price. It is whether the tool reads the document, and that is where Tofu picks up.
None of these tools is wrong, exactly. Dext, AutoEntry, and Hubdoc just draw the line between automated and manual work in different places. Count the line items your team typed by hand last month and the real cost difference shows itself. If you want to see full extraction on your own documents, book a demo and bring a messy one.
None of these three tools is wrong, exactly, they just draw the line between automated and manual work in different places. Your firm's supplier mix, language needs, and document volume should drive the choice more than the monthly price tag. Once you know how many line items you're typing by hand each month, the real cost differences become obvious. When you're curious what full line-item extraction looks like on your own documents, book a demo and bring a messy one.
Neither handles it well. Dext extracts from English and major European languages, and AutoEntry's document guidelines list Arabic, Chinese, and Japanese as unsupported. Invoices from suppliers in Bangkok, Shenzhen, or Dubai fall outside both tools' stated scope.
Hubdoc ships free with Xero, or about $12 a month standalone, because it captures header data only: supplier, date, total, and invoice number. Every line item still gets typed and coded by hand once the document lands in the ledger, at any price point.
AutoEntry charges 1 credit for a standard invoice, 2 if you extract line items, and 3 per bank statement page, and unused credits expire after 90 days. A firm with detailed documents burns through its allowance faster than the plan name suggests. A flat plan does not meter line items or statement pages separately.
No. Dext includes line items only on its higher plans, and AutoEntry charges an extra credit per invoice for them. Tofu extracts every line item by default on every plan, at $79 or $199 a month with no per-document surcharge.
Dext does, with 10 pages included on the Business plan and extra pages billed per page. AutoEntry does, at 3 credits per page. Hubdoc fetches and stores statements automatically but does not extract the transaction lines; for Xero clients the bank feed does that job.
Choose Hubdoc if you need a clean document inbox feeding Xero and your suppliers send simple, English-language invoices without much line-item detail. Once your firm processes multi-line wholesale invoices, non-Latin scripts, or handwritten receipts, header-only capture becomes the limit no matter what you pay.