ReconPe

ReconPe markets a live early-access AI-agent reconciliation product for Indian commerce. It supports marketplace settlement, payment-gateway, bank and rate-card workflows; four named agents; deterministic ACRE matching; human approval; team assignment; approval chains; and audit trails.

ReconPe documents multi-format ingestion, document extraction, AI-assisted schema mapping, cross-run memory, guided investigation, seven Indian format detectors, commission variance review, COD remittance tracking, AI-drafted rules, exception-resolution suggestions, reports and risk prioritization.

The FAQ says ReconPe is live in early access; humans review, approve and close; deep audit is live for Amazon India, Flipkart, Meesho and Razorpay; Cashfree, PhonePe and PayU are detection-only; and paid marketplace plans start at INR 3,999 per month.

Current marketplace pricing lists Free at INR 0, Starter at INR 3,999 monthly or INR 39,990 annually, Growth at INR 6,999 monthly or INR 69,990 annually, and Enterprise as custom. Limits include 5 total free reconciliations, 30 per month on Starter and 100 per month on Growth.

ReconPe says it parses Amazon Settlement Report V2 and MTR data, audits GST-aware fees and category rate cards, matches returns and RTO entries to original orders, and ages COD remittances through bank receipt.

ReconPe says it parses Flipkart commission, fixed, collection, shipping and F-Assured fees; compares each against rate cards; links later returns and reversals to original orders; and handles split settlements.

ReconPe says it parses Meesho Supplier Payments, reconciles supplier-price semantics, tracks COD remittance ageing, carries return deductions across cycles and isolates contestable penalty deductions with order context.

ReconPe documents UTR/RRN matching, per-transaction MDR variance checks, T+1 to T+3 cycle handling and refund/chargeback linkage. The core workflow is file-upload based and does not require a Razorpay API integration.

ReconPe matches GSTR-2B to purchase registers, GSTR-1 to sales registers and GSTR-3B to source returns; flags ITC at risk and unclaimed credit; and exports Excel/PDF close packs. It describes AI-assisted mapping with user confirmation.

FinanceOps is a separate close workspace for AR, AP, bank, intercompany, fixed-asset and inventory sub-ledger-to-GL tie-outs, approvals, period controls, journal-entry drafts, audit packs and CFO reporting.

FinanceOps lists a free one-entity wedge and Ledger at INR 19,999 per entity per month with a two-entity floor, or USD 499 per entity per month globally. The page says six close domains, approvals, close automation, Tally XML export and advisor console are live, while direct NetSuite, QuickBooks and Zoho integrations are building.

Enterprise offers dedicated SaaS or on-prem deployment, custom rules and integrations, SSO/SAML, regional data residency, a dedicated success manager, custom compliance reports and a scoped 60-day pilot. The public process targets pilot kickoff in about three weeks.

ReconPe says AI never decides matches; ACRE is deterministic; every match has per-field evidence; India is the default data-residency location; access roles are enforced server-side; approval chains separate reviewer and approver; and material actions enter an append-only audit log.

An illustrative seven-stage product journey shows file ingestion, AI schema proposals, user mapping approval, deterministic matching, AI investigation and resolution proposals, cross-run linking, human approval/sign-off and an exported close bundle. Displayed metrics are demo figures, not customer outcomes.

ReconPe describes its active test market as Indian commerce and positions the product around fee variance, missing remittances, finance/operations/CA review and repeated month-end reconciliation.

The company reference calls ReconPe live in early access, identifies Indian marketplace sellers, e-commerce finance teams and CAs as the active market, and names Amazon India, Flipkart, Meesho, Razorpay, Cashfree, PhonePe and PayU.

The founder identifies Bengaluru as his ReconPe location and describes a founder-built product, cross-run memory, human approval, multi-format document intelligence, a prompt-failure catalogue and correction logging. A Flipkart demonstration is explicitly described as sample data, not a customer result.

The official company walkthrough shows AI proposing field mappings and rules, user approval before execution, deterministic matching, typed differences and suggested fixes; the user decides what closes.

The official company post demonstrates bank-payout matching, fee-leak detection against a merchant rate card, GST analysis and GMV-to-bank tracing on demo data. The figures are not presented as a named customer outcome.

The official sitemap enumerates product, pricing, marketplace, FinanceOps, learning, FAQ, trust, registration and bookkeeper pages. It contains no customer, case-study, success-story or testimonial URL.

ReconPe operates a curated directory of bookkeepers and CAs covering marketplace reconciliation, GST and month-end close. Engagements are contracted directly; ReconPe states that it does not handle payments or take a per-engagement commission.