01 · LIVE
Reconciliation & exceptions
Core GL against NCHL/ConnectIPS settlement, Tally against bank statements, and more. Five matching stages, AI-suggested matches for review, and every exception arrives with a likely cause.
Every number, accounted for
Verisum sits on top of your core banking system or ERP. It reconciles your accounts, explains every exception in plain language and keeps an audit trail that can’t be altered. Your people approve every decision.
The problem
84%
of finance organisations have implemented AI or plan to.
7%
report a high or very high impact.
The gap is in the method, not the models.
We start where the value can be measured, and build in the controls your auditor and regulator will ask for.
Source: Gartner survey of CFOs
Modules
All modules share one platform: the same connectors, one exceptions workbench and the same governance layer. Reconciliation is live today; the others are being built in this order.
01 · LIVE
Core GL against NCHL/ConnectIPS settlement, Tally against bank statements, and more. Five matching stages, AI-suggested matches for review, and every exception arrives with a likely cause.
02 · IN BUILD
Invoice capture from typed PDFs and spreadsheets, three-way match against purchase orders and receipts, duplicate flags and approval routing.
03 · IN BUILD
Close checklists with owners and deadlines, journal checks, and variance notes drafted for review.
04 · IN BUILD
A 13-week cash forecast and budget against actual, with the drivers behind each number explained.
05 · IN BUILD
Rule and behaviour-based alerts grouped into cases, with an evidence pack ready for the STR process.
06 · IN BUILD
Staging, PD/LGD/EAD inputs and ECL calculation in the NFRS 9 structure, plus early-warning signals with reason codes.
Governance built in
Under Nepal Rastra Bank’s draft AI Guidelines, AI services from an outside vendor need outsourcing due diligence, board approval and notification to NRB. We prepare that file with you.
Your data stays on your premises or on Nepal-hosted servers. Nothing is sent to an outside AI service without your approval.
Every model is registered with an owner, a risk tier and a review date.
Every match and flag shows why it was made, so any decision can be explained.
A named person approves every material action, never the person who proposed it. Any model can be paused at once.
Every action is chained by hash, so any change to history is detected. Exportable for IS audit.
Who it’s for
Class A–D, PSPs and PSOs
Mid-size Nepali companies
SMEs, mid-market, accounting firms
How we engage
2 weeks
Map one process, set the baseline, write the business case.
Gate: go / no-go8–10 weeks
One module on your real data, run alongside your current process.
Gate: KPI target met6–12 weeks
Harden, integrate, train users, hand over, hypercare.
Gate: user sign-offOngoing
Add modules; quarterly model reviews and KPI reporting.
Gate: quarterly reviewOur founder spent 19+ years at the centre of telecom performance, billing migrations, revenue assurance and fraud management. We bring the same discipline to your ledgers, and we share the risk.
Next step
A 45-minute discovery call, an NDA, and one month of sample files, which can be anonymised. You get a baseline, target KPIs and a clear go/no-go.
+977 9851220145
Bhaisepati, Lalitpur-25, Nepal