Onboard · reconcile · underwrite · report
Faster onboarding,under a full audit trail.
Documents read and checked, ledgers matched daily, memos drafted with their sources, reports assembled from the record, and every automated action logged for the auditor who will ask.
Every automated action must be explainable to an auditor.
THE PROBLEM
A new customer sends documents by email, someone checks them against a list, and the file goes back and forth for weeks. Month-end reconciliation is a spreadsheet exercise across bank statements, the ledger and partner files. An underwriter reads bank statements by hand and writes a memo from scratch. Regulatory returns are assembled from several systems by people who know where the numbers are. Circulars are read by the one person who reads circulars.
HOW SOLONOMOUS DOES IT
We automate the preparation and leave the decision with the person who is accountable for it. Documents are read and checked against your rules, and the checker receives a complete file. Reconciliation runs daily and explains each break. Statements are extracted and a memo is drafted with every figure linked to its source. Returns are prepared from the ledger on a schedule and wait for sign-off. Models are evaluated and documented before release, and every action they take is logged with its inputs.
WHAT YOU GET
Onboarding is a review, not a chase. Breaks are found the day they occur. Underwriters read memos instead of writing them. Returns are ready before they are due. The audit trail is complete because it was never optional.
Workflows
What we automate here.
7 workflows
- 01
Customer onboarding and KYC
TODAY
Identity, address and financial documents arrive by email and are checked by hand against a list. Missing or expired items are requested one at a time, and the file is keyed into the core system at the end.
AUTOMATED
Documents are read, checked against your KYC rules and screened, missing items are requested from the customer, and a complete file is placed in the checker's queue. The approval stays with your compliance team.
- 02
Reconciliation
TODAY
Bank statements, the general ledger, payment-gateway files and partner statements are matched in spreadsheets at month end. Breaks are investigated by whoever has time.
AUTOMATED
Feeds are matched daily against the ledger with the rules you set. Each break is classified and routed with the candidate matches shown, and the reconciliation report is ready for sign-off every morning.
- 03
Underwriting support
TODAY
An analyst reads financial and bank statements, computes ratios in a spreadsheet and writes a credit memo from a template.
AUTOMATED
Statements are extracted into structured data, ratios and covenants are computed against your policy, and a memo is drafted with every figure linked to the page it came from. The underwriter reads, questions and decides.
- 04
Customer service
TODAY
Customers ring or write to ask about balances, statements, claims and applications. Agents look things up across several screens while the customer waits.
AUTOMATED
After authentication, an assistant answers account and application questions from the systems of record, by phone, chat and email, in the customer's language. Anything involving money moving or a complaint is handed to an agent with the conversation attached.
- 05
Regulatory reporting
TODAY
Returns are assembled from extracts of several systems by people who know where the numbers live. A late correction means starting the workbook again.
AUTOMATED
Figures are pulled from the ledger and the source systems on a schedule, checked against the prior period and the control totals, and prepared in the return's format for review and sign-off. Every figure traces to its source entry.
- 06
Policy and circular questions
TODAY
Internal policies, product terms and regulator circulars are PDFs in a shared folder. Staff ask the compliance team, who answer the same question again and again.
AUTOMATED
An assistant answers from your own policies, product documents and the circulars you load, with the source paragraph attached. It does not answer from anything outside those documents.
Done with
- 07
Fraud and anomaly review
TODAY
Rules flag transactions, and a queue of alerts is worked by analysts in the order it arrived. Most alerts are closed as false positives after the same checks.
AUTOMATED
A model trained on your own history scores each alert and attaches the checks an analyst would run first. Analysts work the queue in order of risk, and every decision they make is recorded and used to retrain.
Engagement
How it starts in a regulated firm
Compliance and internal audit are in the first meeting, not the last.
- 01
Map one process end to end
We follow one customer or one month-end through every system, spreadsheet and approval and list where documents wait and where decisions are made.
- 02
Agree the control model
With compliance and audit we set what a system may do on its own, what needs a maker and a checker, how every action is logged and where the data may live.
- 03
Run in parallel, evaluated
The automation runs beside the existing process for an agreed period. Its outputs and any models are evaluated against what your people did, and the evidence is written up for your risk team.
- 04
Hand over, or keep running
Your operations and technology teams take it on with the runbooks, code and evaluation results, or we run it under your keys inside your region.
Details
On the spec sheet.
- Systems integrated
- Core banking, lending and policy administration systems, payment gateways, bank statement feeds, KYC and screening providers, general ledgers (SAP, Oracle, Tally and others), CRM, document management
- Data handled
- Identity and financial documents, transactions, ledger entries, statements, applications, credit memos, regulatory returns, policies and circulars
- Compliance considerations
- RBI and SEBI guidelines as they apply to your licence, IRDAI for insurers, KYC and AML rules, an audit trail on every automated action with its inputs, models evaluated and documented before release, data kept in India where required
- Decision boundary
- Credit, KYC approval, claims and money movement are decided by an accountable person; systems prepare, check and record
- Deployment
- In your cloud region or on-premise, inside your network; run by us under your keys where you prefer; no customer data used to train shared models
- Languages
- Customer assistants and voice in the languages your customers speak; memos, returns and reports in English
- Hand-over
- Runbooks, source code, model weights, evaluation reports and the training-data pipeline, with your technology and risk teams trained to operate and re-evaluate them
Questions
Asked before we start.
Can a system approve a customer or a loan?
Not in what we build. Systems read, check, screen, compute and draft; an accountable person approves. Where you want low-risk cases to pass on rules alone, that is a control decision your compliance team makes, and the rule and its outputs are logged.
How is the audit trail kept?
Every automated action is written as an event with its inputs, the rule or model version that produced it, the output and who reviewed it. The log is append-only and can be exported in the form your auditors ask for.
How do you evaluate a model before it touches production?
Against a held-out set drawn from your own history, with the metrics your risk team names and a written report of where it fails. The same evaluation reruns on a schedule after release, and a drop pauses the model.
Where does the data live?
In your own cloud region or your own data centre, under your keys. Nothing is used to train models shared with anyone else. Where a regulation requires data to stay in India, the stack is deployed so that it does.
Our core system is old and has no API. Is that a problem?
It is common. An unattended bot works through the system's screens under a named service account, with every action logged, until an interface is available. Reconciliation and reporting read the extracts the system already produces.
Start
Tell us what your teamstill does by hand.
We map the workflow with you and say plainly what can run on its own, what should not, and what it takes.