Match a ₹25 lakh JE to twelve invoices in seconds.
Last reviewed·ReconPe Editorial
ReconPe FinanceOps brings the ACRE engine to the close. Subset-sum matching for sub-ledger to GL tie-out, Bayesian confidence per group, stateful memory across periods, and an audit trail your Big Four reviewer will accept. Built for the controller closing the same books every month — not the operator chasing a settlement file.
Wave 1 covers AR-to-GL and bank-to-GL with subset-sum matching. Intercompany mirror matching, multi-entity close, and AP-to-GL follow in Q3.
12 AR invoices, summed to one GL journal entry — in seconds, within tolerance.
Pilot data · March 2026 cycle · listed manufacturer · five legal entities
Made for the people who actually close the books
ReconPe FinanceOps is designed by accountants, for accountants. No "data engineering setup." No SQL. No prompt engineering. Just the close — done in the language your team already speaks.
Owns the close
You sign off the trial balance every month. ReconPe gives you AR/AP/Bank-to-GL tie-outs with reviewer/approver stamps and a residual band you can defend.
Runs the workflow
You set up the rule library, manage close calendars, and unblock breaks. ReconPe gives you a CloseCycle dashboard, reusable templates, and per-entity status.
Sees all entities
Multi-entity rollups, intercompany mirror checks, and consolidated discrepancy view. Know which subsidiary is blocked on what — without phoning the controller.
Statutory & tax close
GST/TDS-aware tolerances, GSTR-2B vs GL reconciliation (Wave 2), and per-section workpapers your articleship would recognise. Zero new SQL.
Trusts the trail
Every match is explainable: subset of source rows, residual, confidence, rule, approver, timestamp. Reruns are deterministic. Big-4 review-ready by default.
Governance & SOX
Segregation-of-duties on rule edits and approvals, immutable audit log of dispositions, period-locking, and exportable evidence packs for SOX 302/404.
Speaks the systems your team already uses
ReconPe ingests CSV, Excel, JSON and XML extracts from any source — schema detection recognises common ledger and bank shapes without manual mapping.
Direct connectors are on the roadmap. The export-and-upload flow gets you to a clean tie-out in the first session — no IT ticket, no API keys.
Where the close actually breaks
One JE, twelve invoices
Your GL posts a single ₹25 lakh entry. Your AR file has twelve invoices that should net to it. Spreadsheet matching breaks the moment cardinality stops being one-to-one — and that's most of close.
TDS, GST, FX, and rounding leave residuals
Gross invoice totals rarely match net GL postings to the rupee. Withholding, GST input adjustments, FX revaluation, and bank charges create small variances that hide real breaks under noise.
Bank statements lag, GL doesn't wait
Cash GL closes on the period boundary. Bank statements drift by days. Reconciling them by hand means tracking which transactions belong to which period — across multiple bank accounts and currencies.
Audit asks for a paper trail
Every match needs an explanation. Every break needs a disposition. Every rerun needs to look the same as last time. Excel cannot tell your auditor why two rows tied — but ReconPe can.
What FinanceOps gives you on day one
Subset-sum AR-to-GL matching
ACRE finds the subset of AR invoices that sums to your GL entry within tolerance — currency, counterparty, and date-window aware. Twelve-row matches on a 5,000-line AR aging are seconds, not hours.
Bayesian confidence per match
Every aggregation candidate gets a composite confidence score. You see which groups are safe to approve in bulk, which need a second look, and which residuals are within real-world tolerance for TDS and rounding.
Stateful exception memory
An invoice that didn't settle this month doesn't disappear. ReconPe fingerprints the exception by counterparty, amount, and direction — and links it to the settling entry whenever it lands. Across runs, across periods.
Five named agents do the work — so the close has a vocabulary
Every reference to Iris, Argus, ACRE, Clio, or Hermes in our docs and product UI maps to a specific stage of the close. Same names across the journey, the chat agent, and the audit log.
Iris
Reads CSV / Excel / JSON / XML extracts as-is. Detects field roles (amount, date, counterparty, key) without manual mapping.
Argus
Proposes SUM_AGGREGATE, value-mismatch tolerance, and date-window rules in plain English. Investigates each unmatched break.
ACRE
L1 exact + L2 fuzzy + L3 LSH plus subset-sum aggregation. Twelve invoices to one ₹25 lakh JE in 240 ms.
Clio
Fingerprints every break by counterparty + amount + direction. Auto-links late settlements across runs and learns counterparty patterns.
Hermes
Flips entity period to CLOSED. Exports the audit-ready bundle — matched pairs, residuals, rule reasoning, reviewer/approver stamps.
What close day looks like — before and after
Same trial balance. Same files. Same controller. Different timeline.
- 08:00 — Pull AR aging, GL extract, bank statement into one folder
- 09:30 — Manually map columns; rename headers; clean stray rows
- 11:00 — Filter AR by month; build VLOOKUPs against GL
- 14:00 — Track down the ₹25 lakh JE: which 12 invoices? Try sumifs
- 17:00 — Email vendor team about TDS rounding diffs (every month)
- Day 2 — Re-run on updated bank file. Rebuild some lookups
- Day 3 — Ship to reviewer. Reviewer sends 9 questions. Loop again.
- 08:00 — Drag three files into ReconPe
- 08:04 — Iris detects schemas; field roles assigned
- 08:10 — Argus proposes SUM_AGGREGATE rule; you approve
- 08:13 — ACRE returns 9 aggregation candidates · 240 ms
- 08:25 — You confirm subsets · Memory auto-links 18 late-payers
- 08:42 — Wrong-vendor reversal flagged; you re-post on the GL
- 08:47 — Hermes exports audit-ready bundle. Period CLOSED.
Compared to what your team uses today
Honest read of where each tool actually breaks. We don't claim to replace your ERP — we replace the spreadsheet that lives between your sub-ledger and your GL.
| Capability | Excel + macros | Generic recon tool | RPA bots | ReconPe FinanceOps |
|---|---|---|---|---|
| Subset-sum many-to-one matching | VLOOKUP breaks at >1:1 | 1:1 only | Mimics clicks, no math | Native ACRE aggregation |
| Residual tolerance for TDS / GST / FX | Manual nudge per row | Flat tolerance | Hardcoded thresholds | Configurable + counterparty-aware |
| Cross-period exception memory | Lost when file closes | Reset per run | No state model | 180-day fingerprinted pool |
| Audit trail for Big-4 review | Reconstruct from emails | Match log only | Bot run history | Subset · residual · rule · approver |
| Time to first reconciliation | Hours of cleanup | Schema mapping project | Weeks of bot scripting | Drag-drop, minutes |
| Deterministic reruns | Depends on the analyst | Mostly | Brittle to source layout | Same inputs, same outputs |
Side-by-side reads from pilot deployments March 2026. We update this table when our feature set changes; we don't update it when a competitor's marketing does.
Week 1 with ReconPe FinanceOps
We don't ask you to migrate everything on day one. We ask you to replay last month's close on real data. If the numbers don't tie, you walk away.
- Day 1Map your close calendar
We sit with your controller, list every reconciliation in your trial balance, and pick the 2–3 highest-pain ones to migrate first.
- Day 2Upload real extracts
AR aging, GL, bank statements from your last closed period. Iris fingerprints columns; you validate the schema in 10 minutes.
- Day 3Build & approve rules
Argus proposes rules for each tie-out — SUM_AGGREGATE, value-mismatch tolerance, date-window. You review and approve in plain English.
- Day 4Re-run last close
Replay your last month's close on ReconPe end-to-end. Compare auto-match rate, residuals, and time-to-close against your spreadsheet.
- Day 5Ship live close
Cut over to ReconPe for the next live period. Memory carries forward. By month 3, your close runs in under an hour with the same headcount.
Security, controls, and the audit trail by default
Your statutory auditor and your audit committee shouldn't have to take our word for it. Every match is reconstructable and every override is logged.
SOX 302 / 404 evidence pack
Per-period exportable bundle: matched pairs, residuals, rule reasoning, reviewer + approver stamps. Hand it to your auditor as-is.
Reviewer / Approver SoD
Segregation-of-duties enforced on rule edits and exception dispositions. Same user can't propose and approve.
Period locking
Once a CloseCycle flips to CLOSED, the period is read-only. Reopen requires an explicit role + audit reason.
Immutable audit log
Every match, override, rule change, and reopen is appended to a tamper-evident log with user, timestamp, and reason.
180-day exception memory
Per-organisation fingerprint history with aging. Answer 'why did we close this break in March?' in October.
What lands next — the multi-entity and statutory wedges
We don't promise these in the Wave-1 contract. They're on the roadmap with named quarters, and we'll only call them shipped when they actually are.
Intercompany mirror matching
Entity A's debit must mirror Entity B's credit. Asymmetric postings flagged as invariant violations rather than silent breaks.
Multi-entity CloseCycle dashboard
Roll up tie-out status across legal entities. See which subsidiary is blocked on what, without phoning the controller.
GSTR-2B vs GL reconciliation
Rate-card-aware GST validation. Input tax credit reconciliation against the books, native to Indian filers.
Direct ERP connectors
Tally Prime, Zoho Books, NetSuite, SAP S/4HANA. Cuts the export-and-upload step out of the period cadence.
What controllers, CAs, and auditors actually said
Pilot quotes — paraphrased and approved by the speakers. Names withheld until general availability; designations and contexts are real.
"Three days of reconciliation became ninety minutes — and the audit committee gets a cleaner trail than what we used to ship in our workpapers."
"TDS rounding diffs used to be a parking lot of three-page reconciliation notes. Now they sit inside the matched group with a residual flag — and the partner signs off without asking us to redo the workings."
"I don't get phone calls about which subsidiary is blocked anymore. The CloseCycle dashboard tells me, and the controller and I see the same screen. That alone is worth the seat licences."
Frequently asked
Can ReconPe match many AR transactions to one GL journal entry?
Yes — that is the FinanceOps wedge. ACRE's aggregation stage searches the unmatched AR pool for subsets whose amounts sum to a target GL entry within configurable tolerance. Currency, counterparty, and date-window pruning keep the search bounded. The default tolerance absorbs TDS withholding, GST adjustments, and rounding so real-world residuals don't block automatic matches.
How is FinanceOps different from ReconPe's marketplace product?
Same engine, same memory, same agent modes — different starting position. FinanceOps leads with sub-ledger to GL tie-out (AR-to-GL, AP-to-GL, bank-to-GL, intercompany mirror matching), pre-built rule templates for ledger formats, and a Ledger pricing tier billed per legal entity rather than per reconciliation. If you close the same books every month, FinanceOps is the entry point.
Does it work with Tally, Zoho Books, QuickBooks, NetSuite, or SAP?
ReconPe ingests CSV, Excel, JSON, and XML extracts from any source system. The schema detection layer recognises common GL extract shapes (journal_id, posting_date, debit, credit, account) and AR aging shapes without manual column mapping. Direct connectors are on the roadmap; for now an export-and-upload flow gets you to a clean reconciliation in the first session.
How does FinanceOps handle multi-entity close?
Each legal entity gets its own period workspace. Reconciliations roll up into a CloseCycle dashboard so you can see which entities are tie-out pending and which are blocked on a specific exception. Intercompany mirror matching enforces the invariant that Entity A's debit must mirror Entity B's credit, and asymmetric postings are flagged as invariant violations rather than silent breaks.
Is the audit trail strong enough for a Big Four review?
Every aggregation match is explainable: the subset of source rows, the residual, the confidence score, the rule that triggered, and the user (or agent) who approved the disposition. Reruns are deterministic given the same inputs. The exception memory pool keeps a per-organisation fingerprint history with 180-day aging so you can answer 'why did we close this break in March?' in October.
What about GSTR-2B reconciliation against the GL?
GSTR-2B versus GL tie-out is on the FinanceOps roadmap as a dedicated wedge for Indian filers. It is not a connector ReconPe ships today — uploading the GSTR-2B JSON or Excel export and a GL extract produces a reconcilable surface, but the rate-card-aware GST validation (input tax credit reconciliation against the books) is a Wave 2 capability. We don't promise it before it ships.
Close the books. Not the ticket queue.
FinanceOps early preview — request access and we'll set up your first AR-to-GL reconciliation against your own data within a week.
Billed per legal entity, not per reconciliation. Run as many tie-outs as your close calendar needs.