Patents and research at Terra Insight
Terra Insight runs a small, India-anchored research practice. Our work is grounded in real customer statement corpora, real Indian regulatory texts, and the operational reality of running reconciliation infrastructure at enterprise scale. This page covers our patent posture and the research artefacts we publish openly.
Variance taxonomy framework for reconciliation
Terra Insight has filed a patent on its variance taxonomy framework — the structured classification of every reconciliation variance into a named, recoverable category. The taxonomy is the layer that turns "the books don't match the bank" into an actionable list: this many rupees are stuck in timing differences, this many are TDS reclassifications, this many are statutory deductions still pending recovery, this many are platform fee decompositions that need a credit note.
The taxonomy is what makes TransactIG audit-ready by default. Each variance carries evidence, a category, and a recovery path — not just a difference figure. We have chosen not to publish the application number on the public site; reviewers and procurement teams can request the reference under NDA.
India-specific data, customer corpora, no synthetic benchmarking
Our research practice has three commitments. The first is India-specific data primacy. We do not borrow models or evaluation sets from US or European reconciliation vendors. Indian bank narrations, NACH return codes, GST tax structures, TDS sections, and platform settlement formats are not edge cases we adapt to — they are the design surface from day one. Every preset, every parser, every taxonomy entry is built against Indian source documents.
The second is customer-statement-corpus learning. The improvements we ship come from working through real customer bank statements, real ERP exports, and real settlement files — with consent, under our ISO 27001:2022 data-handling controls. This is the only way to make progress on the messy long tail: PSU bank narration patterns, co-op bank statement quirks, marketplace fee schedules that change quarterly, NBFC mandate mismatches. None of this surfaces in clean public datasets.
The third is a refusal to publish synthetic benchmarks. We do not run our internal accuracy numbers against generated or scrubbed statement data and call it a benchmark. The numbers we publish — like the 51% to 88% match-rate improvement on a customer engagement — are real outcomes on real corpora. We would rather publish one honest customer outcome than ten manufactured benchmarks.
What we publish openly
Founder profile — operating background in supply chain finance, sales operations, and enterprise data at Amazon India, Intuit, and the Tata Group.
India-specific tier-1/tier-2 reconciliation pattern: RMPV, ITC-04, multi-plant settlement matching.
Browser-based calculators covering RMPV variance, ITC-04 tracking, FASTag, steel yield, GST rate lookup, and three-way-match cost.
Full coverage of all 93 controls across the four 2022 themes, including the 11 new controls introduced in the revision.
Encryption, access control, audit logging, tenant isolation, and DPDP Act 2023 alignment — applied across both TransactIG and TransactIQ.
850+ published research notes on Indian TDS, GST, NACH, banking, and platform settlement reconciliation.
Talk to the team
Reviewers, procurement teams, and prospective customers can request our patent reference, ISO 27001:2022 certificate, and security documentation under NDA.