llms.txt
Compare with other sites' llms.txt files →# Cointab Context Document - New Reconciliation Platform
Company: Cointab
Product: Financial reconciliation automation platform
## 1. One-line description
Cointab is an AI-assisted reconciliation platform that helps finance teams match internal records with external records, identify discrepancies, review matched/unmatched transactions, and download audit-ready reconciliation reports.
## 2. What problem Cointab solves
Finance teams spend a lot of time reconciling data across multiple systems such as sales reports, books, ERP exports, payment gateways, marketplaces, bank statements, logistics partners, vendors, customers, and tax records.
Most teams still do this using Excel, formulas, VLOOKUPs, pivot tables, manual checks, and repeated file comparisons. This creates several problems:
- Reconciliation is slow and repetitive.
- Excel formulas break or become hard to audit.
- Different team members prepare reports differently.
- Large files become difficult to handle.
- Exceptions remain open for too long.
- Missing payments, deductions, refunds, returns, fees, or settlement differences may go unnoticed.
- Month-end close and audit preparation become stressful.
- The same reconciliation setup is repeated again and again for different customers or periods.
Cointab replaces this with a structured, reusable reconciliation workflow where users upload data, map required fields, run reconciliation, review results, and export reports.
## 3. Core product positioning
Cointab is not just a bank reconciliation tool or payment gateway reconciliation tool. It is a flexible reconciliation engine for comparing any two sides of financial or operational data.
It can be used for:
- Sales vs payment gateway reconciliation
- Marketplace sales vs settlement reconciliation
- eCommerce order vs payment reconciliation
- COD delivery partner reconciliation
- Bank statement vs books reconciliation
- Vendor reconciliation
- Customer reconciliation
- PSP / payment gateway reconciliation
- Logistics / freight invoice reconciliation
- Marketplace reconciliation
- Intercompany reconciliation
- Tax / statutory data reconciliation
- Any custom internal vs external data reconciliation
The core idea is:
Side A = records the business expects to be correct
Side B = records received from external systems, partners, banks, marketplaces, PSPs, or other sources
## 4. Who Cointab is for
### Target companies
Cointab is useful for companies that deal with high-volume or multi-source transaction data, especially where money, settlements, invoices, orders, payments, or statements need to be matched regularly.
Ideal company types include:
- eCommerce brands
- D2C brands
- Online sellers
- Retail companies
- QSR / restaurants
- SaaS companies
- Marketplaces
- Logistics companies
- Payment-heavy businesses
- Fintech companies
- Agencies and professional service firms
- Accounting / CA / CPA firms
- Mid-market finance teams
- Companies with multiple PSPs, banks, marketplaces, delivery partners, or vendors
### Target users
Cointab is mainly used by:
- Finance teams
- Accounts teams
- Controllers
- CFOs
- Founders of lean companies
- Reconciliation analysts
- Accounts receivable teams
- Accounts payable teams
- Marketplace operations teams
- eCommerce operations teams
- Audit and compliance teams
- Accounting firms handling reconciliation for multiple clients
## 5. Customer value
Cointab helps customers get value in the following ways:
### Time savings
Cointab reduces the time spent on repetitive reconciliation work by automating data preparation, matching, exception identification, and report generation.
### Better accuracy
The system applies structured matching logic consistently, reducing manual errors caused by Excel formulas, copy-paste mistakes, and inconsistent review methods.
### Reusable setup
Once a reconciliation is configured, it can be reused for future periods. Users do not need to rebuild the same reconciliation repeatedly.
### Recurring automation
Once a reconciliation is configured, Cointab can automate data collection, reconciliation runs, and output delivery using email, SFTP, or API integrations. This reduces daily manual work and helps customers keep internal finance, accounting, analytics, and BI systems updated automatically.
### Faster exception handling
Cointab separates fully matched, partially matched, unmatched, and skipped transactions clearly. This helps finance teams focus only on exceptions instead of reviewing every transaction manually.
### Audit-ready reporting
Users can download Excel reconciliation reports containing matched, partially matched, unmatched, and skipped records for internal review, audit, and partner follow-up.
### AI support
AI helps users create derived columns using natural language, analyze difficult open transactions, and identify possible reasons and actions for unresolved items.
### Team collaboration
Cointab supports team-based workspaces where multiple users can work under one common team account with roles and access control.
## 6. Main system flow
Use this flow:
1. User signs up.
2. User enters a team workspace.
3. User starts a new reconciliation.
4. User either selects a popular reconciliation or creates a custom reconciliation.
5. User uploads required files on Side A and Side B, or configures automated data input.
6. User maps required fields such as date, amount, and identifiers.
7. User optionally uploads supporting data for lookups, merging, enrichment, or calculation.
8. User optionally creates derived columns using AI-generated Excel formulas.
9. User runs reconciliation manually, or schedules it to run automatically.
10. System performs structured matching using the reconciliation engine.
11. Open transactions are further analyzed using AI where rules are not sufficient.
12. User sees live progress while reconciliation is running.
13. Once completed, user reviews the reconciliation report.
14. User can explore matched, partially matched, unmatched, and skipped transactions.
15. User can use filters for deeper analysis.
16. User can download the Excel report.
17. Cointab can optionally push the reconciliation output back to internal, accounting, analytics, BI, or other systems through email, SFTP, or API.
18. User can manually match transactions that the system and AI could not match.
19. Reconciliation remains available on the dashboard for future reference.
20. If a file was missed, user can upload the missed file under the same reconciliation and refresh the report.
21. The same reconciliation setup and automation can be reused for future periods.
## 7. Reconciliation types
Cointab supports two broad types of reconciliation setup:
## 7.1 Popular reconciliations
Popular reconciliations are pre-built reconciliation templates created by Cointab and made available to customers.
Examples:
- Myntra Sales vs Payment
- Flipkart Sales vs Payment
- Amazon MTR vs Disbursement
- Sales vs PG
- Bank vs Books
- Marketplace vs Settlement
- COD Delivery Partner vs Sales
These are useful when both sides of the reconciliation are based on standard external partner reports. For example, Amazon, Flipkart, Myntra, PSPs, and delivery partners usually provide the same report structure across customers.
In a popular reconciliation:
- Cointab already defines the required file formats.
- Cointab already defines data preparation logic.
- Cointab already defines matching logic.
- User only uploads the required files.
- User selects the period.
- User runs reconciliation.
Popular reconciliations should use a global Cointab-maintained configuration rather than copying the configuration locally for every customer. If the global configuration changes in the future, Cointab can either create a new version or apply the change prospectively. Historical reports should not be silently changed.
## 7.2 Custom reconciliations
Custom reconciliations are created by users for their own business-specific reconciliation workflows.
In a custom reconciliation:
- User defines Side A reports.
- User defines Side B reports.
- User maps date, amount, and identifier columns.
- User can upload multiple reports on both sides.
- User can upload supporting datasets.
- User can create derived columns.
- User can define or use matching rules.
- Once created, the custom reconciliation can be reused for future runs.
Examples:
- Internal Sales Report vs Cashfree + Razorpay + Stripe
- ERP Sales vs Marketplace Settlement
- Books vs Bank Statement
- Vendor Ledger vs Vendor Statement
- Order Report vs COD Delivery Partner Reports
## 8. Side A and Side B model
## Side A - Your Records
Side A contains records the business expects to be correct.
Examples:
- Sales report
- Books data
- ERP export
- Internal order report
- Ledger report
- Customer receivable report
- Vendor payable report
- Internal settlement working
- Source-of-truth business data
Side A is shown on the left side of the reconciliation setup screen.
## Side B - External Records
Side B contains records received from external systems, partners, banks, marketplaces, PSPs, or other parties.
Examples:
- Payment gateway report
- Marketplace settlement report
- Bank statement
- Delivery partner COD report
- Vendor statement
- Customer statement
- Tax authority report
- PSP payout report
- Refund report
- Reverse shipping report
Side B is shown on the right side of the reconciliation setup screen.
## 9. File upload and configuration
Users can upload CSV, XLS, or XLSX files.
For every primary report, user selects:
- Header row
- Date column
- Amount column
- Reference / identifier column or columns
Identifier columns can include:
- Order ID
- Transaction ID
- Invoice number
- Payment reference
- Bank UTR
- AWB number
- Settlement ID
- SKU
- Customer/vendor code
- Any other business identifier
Users can upload multiple files under the same report if they follow the same configured format.
If a file does not match the configured format, the system should reject it with a clear error, such as:
"This file doesn't match the configured format. Missing column: Transaction Date."
## 10. Supporting data
Supporting data is optional.
Supporting data is not reconciled directly. It is used to enrich, complete, calculate, or prepare the primary data before reconciliation.
Examples:
- Product master
- Fee rate file
- Return report
- Order metadata
- Marketplace mapping file
- GST / tax mapping file
- Delivery partner reference file
- Customer/vendor master
- SKU mapping
- Store mapping
- Report needed for VLOOKUP-style enrichment
Supporting data can be uploaded on both Side A and Side B.
Typical use cases:
- Add missing order details
- Merge reports before reconciliation
- Lookup fee or tax rate
- Pull order status from another file
- Add AWB number to sales report
- Combine sales and returns
- Convert partner-specific IDs into internal IDs
## 11. Derived / calculated columns
Users can create derived columns on both Side A and Side B.
A derived column is a calculated column created from existing columns.
Users can create derived columns using AI. The user describes what they want in natural language, and AI generates an Excel-style formula.
Example prompt:
"If Order processing status is Delivered, use Payment Amount. Otherwise 0."
Example generated formula:
=IF(UPPER(TRIM(C1))="DELIVERED",D1,0)
Derived columns can be used as:
- Amount columns
- Identifier columns
- Lookup fields
- Matching fields
- Output fields
Examples of derived columns:
- Clean Order ID
- Net Amount
- Delivered Payment Amount
- Normalized Transaction ID
- Amount excluding tax
- Amount after fee
- Refund amount as negative
- Combined reference
- Clean AWB number
Derived columns are recalculated whenever reconciliation is run.
## 12. Reconciliation engine
Cointab's reconciliation engine uses exhaustive structured logic to match records across sides.
The system supports matching across:
- One-to-one
- One-to-many
- Many-to-one
- Many-to-many
- Net-to-net
- Contra matching
- Partial matching
The system supports identifier logic such as:
- One Side A identifier vs one Side B identifier
- One Side A identifier vs all Side B identifiers
- All Side A identifiers vs one Side B identifier
- All Side A identifiers vs all Side B identifiers
- Side-only contra matching
- Cross-side matching
The system supports comparison methods such as:
- Equals
- Contains
- Similar
- Equals subset
- Contains subset
- Similar subset
The system can match records even when:
- One transaction on one side maps to multiple transactions on the other side.
- Multiple transactions on one side map to one transaction on the other side.
- Multiple transactions on both sides need to be grouped and compared.
- Identifiers are present across different fields.
- Identifiers partially match.
- Amounts need to be netted.
- Contra entries exist.
- Partial amount differences remain.
The engine first applies structured rules. After that, remaining open transactions are further analyzed using AI.
## 13. Role of AI in reconciliation
AI is used in three main ways:
## 13.1 AI formula builder
AI helps users create Excel-style formulas for derived columns.
This is useful when finance users know the business logic but do not want to write formulas manually.
## 13.2 AI-assisted open-item matching
After structured matching is complete, AI analyzes remaining open transactions where deterministic rules may not be enough.
AI can help with:
- Unstructured references
- Slightly different descriptions
- Inconsistent partner data
- Missing or partial identifiers
- Complex grouping scenarios
- Exception transactions
- Business context that is not obvious from simple rules
AI should be conservative and audit-friendly. If evidence is not strong enough, the transaction should remain unmatched rather than creating a weak match.
## 13.3 AI reason and action analysis
For open transactions, AI can help identify:
- Why the transaction may be unmatched
- What action the user should take
- Whether a file may be missing
- Whether a refund, return, fee, delay, or deduction may explain the difference
- Whether the partner should be contacted
- Whether internal records need correction
## 14. Reconciliation run
When user clicks Run reconciliation, the system starts processing the uploaded files.
Depending on data volume, reconciliation may take a few minutes.
The user is shown live output as reconciliation progresses.
The system should communicate progress clearly so the user knows that reconciliation is running and not stuck.
## 15. Reconciliation report
Once reconciliation is completed, user sees a report dashboard.
The report includes:
- Total summary
- Fully matched summary
- Partially matched summary
- Unmatched summary
- Skipped summary
- Transaction-level tables
- Filters
- Detailed matched transaction view
- Download report option
## 15.1 Fully matched
Fully matched transactions are transactions where identifiers and amounts match according to the reconciliation logic.
Example:
Sales transaction matches PG transaction by order ID and amount.
## 15.2 Partially matched
Partially matched transactions are transactions where identifiers match, but amounts do not match.
Example:
Sales order ID matches payment gateway reference, but sales amount is 669.33 and received amount is 668.00.
Partial matches are important because they show that the transaction is probably related, but the difference needs review.
## 15.3 Unmatched
Unmatched transactions are transactions present on one side but not found on the other side.
Examples:
- Present in Sales but not in PSP
- Present in PSP but not in Sales
- Present in Books but not in Bank
- Present in Bank but not in Books
- Present in Marketplace but not in internal records
## 15.4 Skipped
Skipped transactions are records that were not included in reconciliation.
Possible reasons:
- Excluded by rule
- Missing required data
- Invalid amount
- Incomplete record
- Duplicate or unusable row
- File/data issue
Skipped records should be visible so the user understands what was ignored and why.
## 16. Manual match
Cointab provides a manual match option for transactions that could not be matched by the system or AI.
Users can manually select transactions from both sides and match them if the totals tally.
Manual match is useful when:
- The user knows the business context.
- The partner data is incomplete.
- Identifiers are missing.
- AI cannot confidently match.
- A one-off exception needs manual treatment.
Manual matches should be clearly marked as manual.
Users should be able to undo manual matches.
If a transaction was previously part of a partial match and the user manually matches it, the system can break the previous partial match group and create the new manual match.
## 17. Missed file upload and report refresh
If the user forgot to upload a file, they can upload the missed file under the same reconciliation.
After the new file is uploaded, the report should be refreshed.
This is important because in real finance operations, users often receive late reports from PSPs, marketplaces, delivery partners, or banks.
## 18. Period handling
Cointab supports flexible period handling.
Users can reconcile:
- Monthly periods
- Quarterly periods
- Yearly periods
- Lifetime / all-time periods
- Custom periods
Users can also upload subsequent period files under a common period if they want to maintain a running report.
Open items can also be moved or carried forward to the next period where needed.
Examples:
- March 2026
- Q1 2026
- FY 2025-26
- Lifetime
- All_V1
- Custom settlement period
## 19. Reuse of reconciliations
A major value of Cointab is reuse.
Once a popular or custom reconciliation is configured, users should not need to configure it again.
For future runs, they should only:
- Select reconciliation
- Select period
- Upload required files
- Run reconciliation
- Review report
This reduces setup time and prevents repeated configuration mistakes.
## 20. Data automation and integrations
Once a reconciliation is created and configured, users can automate the data flow so that files or data do not need to be uploaded manually every time.
Cointab supports data automation through:
- Email
- SFTP
- API integrations
This allows data to be pushed into Cointab or pulled by Cointab from external systems on a scheduled frequency.
### 20.1 Automated data input
After a reconciliation is built, users can configure data automation for the required reports.
For example:
- A payment gateway report can be received daily by email.
- A marketplace settlement report can be pulled from SFTP.
- A bank statement can be fetched through API.
- An internal sales report can be pushed from the customer's ERP.
- A logistics partner report can be uploaded automatically to Cointab through SFTP.
- A custom system can send data to Cointab through API.
Once data is received or pulled, Cointab loads it into the correct reconciliation workflow.
### 20.2 Scheduled reconciliation runs
Users can define a schedule for automated reconciliation.
Examples:
- Daily
- Weekly
- Monthly
- End of day
- Every morning
- After file receipt
- After all required files are received
- Custom scheduled frequency
Once the required data is available, Cointab can automatically start the reconciliation without manual user action.
Typical automated flow:
1. Cointab receives or pulls the required data.
2. Cointab validates the file/data format.
3. Cointab loads the data into the correct reconciliation.
4. Cointab runs reconciliation automatically.
5. Cointab prepares the reconciliation report.
6. User is notified when the report is ready.
7. Output can optionally be pushed back to other systems.
### 20.3 Automated output delivery
After reconciliation is completed, Cointab can also push the output back to other systems.
Output can be delivered through:
- Email
- SFTP
- API
This helps customers update their internal systems automatically.
Examples of