Two files that should match, and almost never do on the first try. Reconciling means comparing line by line. Differences hide in the details, and finding them by hand in spreadsheets takes hours and leaves little record of what was decided.
From mess to match, with the math in view.
A Receipts1 row
10/14TRF 8821348,300.00
Total A48,300.00
B Invoices12 rows
10/02FC 0001-44104,025.00
10/03FC 0001-44113,980.00
10/05FC 0001-44154,112.50
… 9 more invoices
Total B48,300.00
1:N groupDifference 0.00Pending review
Illustrative example
01
The same payment, on another date.
The bank credits it one day; the general ledger records it the day before.
Oct 3 · 1,137.50Oct 2 · 1,137.50
02
One receipt that settles twelve invoices.
A single transfer on one side, many entries on the other.
1 row · 48,300.0012 rows · 48,300.00
03
The same columns, with other names.
Every system exports its own way.
AmountDebit – Credit
The idea
DueHub Match organizes the work. The decisions stay yours.
Two sources.
Proposed matches.
Review the differences.
Decide with clarity.
Silent animation: two columns of strips, fuchsia and turquoise; strips of the same length join into ordered pairs and the ones that do not match move to a tray, where they are marked one by one.
Two sources
Matches
Differences
Decision
How it works
Four steps, in the real app.
Silent video with real DueHub Match screenshots: uploading two files, choosing columns and rules, running the reconciliation, reviewing a group and downloading the result.
Review and control
Every group, explained. Every decision, recorded.
01Totals at a glance
Coverage by amount, balance difference, pending reviews and what remains unreconciled in A and in B.
02Group detail
Each source's entries, their totals and the difference. If a group is unusually large, the app warns you before you approve it.
03Human decision
Approve or reject each group, or several at once. Filter by status and search by document, reference or customer.PendingApprovedRejected
04History
Who built or decided each group and when, within the same reconciliation.
05Honest states
A run can end complete, without an optimality guarantee, or incomplete. The app says so and explains which rows are affected.
App screenshots inTap a screenshot to see it full size.
Illustrative image
Export
Take the result wherever your work continues.
.xlsxSummary, reconciled, unreconciled from A and from B, and audit.
.csvRows from A and B aligned by group.
diag.For support: parameters and timings, without your data.
Use cases
When two lists need to tie out and they don't.
Bank versus general ledger
The month's statement against the ledger entries, with dates that shift by a day or two.
Statement→Ledger1:1
Receipts versus invoices
One receipt that settles several invoices, or several transfers that pay a single invoice.
Receipts→Invoices1:NN:1
Payments versus orders
What left the bank against the payment orders issued, to find what is missing on either side.
Payments→OrdersN:N
Two systems, one period
The ERP export against the team's control sheet, each with its own columns.
ERP→Control sheetTemplate
What each file needs
CSV or XLSX, with a header row. Save the mapping as a template for next time.
Amountrequired
Daterequired
Document or referenceoptional
Customer · account or segmentoptional
In numbers
What the engine does, no fine print.
4group types
1:1, 1:N, N:1 and N:N in the same run.
50,000rows per file
On the Growth plan, per source.
2sources, never mixed
A and B are uploaded, mapped and reviewed separately.
3honest states
Complete, without optimality guarantee, or incomplete.
0of your data in diagnostics
The support file carries parameters and timings.
ES · ENlanguages
The app, in neutral Spanish and English.
4group types
1:1, 1:N, N:1 and N:N in the same run.
50,000rows per file
On the Growth plan, per source.
2sources, never mixed
A and B are uploaded, mapped and reviewed separately.
3honest states
Complete, without optimality guarantee, or incomplete.
0of your data in diagnostics
The support file carries parameters and timings.
ES · ENlanguages
The app, in neutral Spanish and English.
Trust
Clear limits, separate data.
Data per organization
Uploads, reconciliations, templates and results are only queried within your session's active organization.
The two sources never mix
A and B are uploaded, reviewed and mapped separately, from preview to result.
Traceability
History, review decisions and the state of each run stay linked to the reconciliation.
Visible limits
See how many reconciliations you have left this month and how many rows your plan allows before uploading.
Preview · Assistants
Reconcile from Claude or ChatGPT.
Preview available today in Claude and ChatGPT; connector acceptance is still pending. Connect DueHub Match to your assistant. Give it the two files, it shows you the same form as the app to choose columns and rules, and explains the result. The decisions stay yours.
# In Claude: Settings → Connectors → AddURL https://mcp.duehub.app/mcpAuth OAuth 2.1 · each person with their account
REST API · beta
# Upload source A of a reconciliationcurl -X POST https://match.duehub.app/api/v1/commercial/uploads \
-H "Authorization: Bearer $DUEHUB_TOKEN" \
-F side=A -F "file=@extracto.csv"
Plans
Start free, grow when you need to.
Launch pricing, in US dollars. Pay yearly and get 2 months free.
For higher volume or an SLA, the Enterprise plan is quoted on request. Need more in a busy month? Add 25 reconciliations for USD 19.
What's new
The Claude and ChatGPT form, with the app design, in light and darkAssistants
Prioritize the document number: what doesn't match pairs by date and amount and is flaggedRules
Faster many-to-many (N:N), with date-range searchEngine
The app, in neutral Spanish and in EnglishApp
FAQ
What matters before you start.
Which files can I upload?
CSV or XLSX, one per source. In an XLSX you choose the sheet and the header row. The number of rows per file depends on your plan: 500 on Free, 10,000 on Starter and 50,000 on Growth. In large files, 1:1 matching covers the whole ledger and many-to-many (N:N) groups are searched in segments; the app reports what was left out of that search.
Which columns does each file need?
Amount and date are required. Document, reference, customer and account or segment are optional and help tell similar entries apart. The app suggests each column by its name and content; you can always change it.
Which rules can I set?
The accepted date difference in days, the accepted amount difference, an expected difference per group (for example, a fixed fee), the group types to look for (1:1, 1:N, N:1 and N:N), grouping by month and requiring the same document number.
Is the result always complete?
Not always. A run can end complete, complete without an optimality guarantee, or incomplete if some rows were left unexamined. The app shows the state and explains which rows are affected so you can review them.
Can I review before exporting?
Yes. Each group can be opened, approved or rejected, and unreconciled rows from A and B stay listed for follow-up. Then you download the result in Excel or CSV.
Can I use it from Claude or ChatGPT?
Yes, with the DueHub Match connector and manual API/MCP activation for your organization. Each person signs in with their account, and the assistant shows the form to choose columns and rules before launching.
Do I have to set everything up every time?
No. Save the column mapping as a template and reuse it with files in the same format.
How do I get started?
Create an account, upload your two files and follow the steps. The Free plan is enough to try the full flow with your files.
Start today
Less manual searching. More time to decide.
Upload your two files and see what matches and what doesn't. Small files take minutes; large ones are analyzed in segments, with a record of what was covered.
Assistant · DueHub Match connectedexample