Onboarding is a document relay
Identity, address, income and banking documents collected, read, queried and re-collected. Every incomplete file adds days, and the customer is comparing your turnaround with a competitor's app.
Industries / BFSI
BFSI, run by systems that finish the work.
Onboarding that clears the document queue, credit files ready for a decision, and every step reconstructible for an auditor.
KYC document intelligence, fraud signals and compliance-grade assistants.
The picture on the ground
What BFSI actually looks like from the inside.
Lending and insurance run on documents. Identity, address, bank statements, income tax returns, GST filings, property papers, medical reports — each one read by a person, each rejection restarting a clock the customer is counting. Turnaround time is the competitive variable in most retail and MSME products, and almost all of it is spent on reading rather than on deciding.
That makes document intelligence the obvious entry point, and it also makes governance the hard part. A regulated entity remains accountable for every outsourced process, so anything we build has to be explainable, logged, reversible and reviewable. We put extraction, validation and file assembly under automation, and we leave sanction, pricing, rejection and collections treatment with your people.
The same pattern extends to the contact centre and collections. Staff answering product, eligibility and grievance questions need answers grounded in current circulars, not in a colleague's recollection. Early-bucket collections need contact at scale within the timing and conduct rules that fair practice codes set. Both are assistive systems with a person carrying the decision.
Where it hurts
Four pressures we hear in every review meeting.
Named the way your operators name them, not the way a vendor deck names them. If none of these land, we are the wrong partner for this problem.
Underwriters spend their time reading
Bank statements spread by hand, GST returns cross-checked line by line, bureau reports interpreted from scratch. The judgement that requires an underwriter takes a fraction of the time the preparation does.
Collections capacity is fixed while the book grows
Early buckets get generic treatment because there is no capacity for anything else, and accounts that only needed a reminder roll forward into buckets that need a field visit.
Every control has to be evidenced
An auditor or a regulator asks why a specific decision was made on a specific day. The answer has to be reconstructible from records, not assembled from the memory of whoever was on the desk.
What we build
The work that pays for itself first.
Ordered the way we would sequence them in a BFSI engagement. Each one links to the capability behind it.
Onboarding document intelligence
Extraction and cross-validation across identity, address, income and banking documents, with mismatches, poor scans and expired documents flagged at upload. The customer fixes one thing once instead of three times over a week.
Credit files assembled for the underwriter
An agent that pulls bureau, banking, GST and internal history into a structured file and writes a summary with every figure traced to its source document. The underwriter reviews and decides; the system never scores or sanctions.
A product and policy assistant for staff
Branch and contact-centre answers grounded in current circulars, product notes and process manuals, with the clause cited and the version dated. When a circular is superseded, the answer changes with it.
Fraud and anomaly signals, scoped properly
Which signals are worth modelling, what a false positive actually costs your operation, what the monitoring and challenger regime looks like, and how the model is governed and documented before anything reaches production.
Early-bucket collections contact
Reminders and payment-arrangement conversations for accounts a few days past due, within the calling windows and conduct rules that fair practice codes require, with a full log and an immediate route to a human on any dispute or hardship signal.
Deployment inside your own perimeter
Open-weight models served in your VPC or data centre, with no data leaving the environment, plus the monitoring, access control, versioning and audit logging that a technology risk review will ask for.
The systems it plugs into.
Read-only to begin with, write access only where a workflow needs it, and every integration documented before it goes near production.
Core banking or policy administration systemLoan origination and management systemsCKYC records and credit bureau responsesBank statements, GST returns and tax documents
Transaction monitoring and AML alertsContact centre telephony and collections diallerCRM and grievance redressal logsData warehouse and regulatory reporting extracts
How it gets judged
Agreed before the build, measured after it.
These are targets and mechanisms, not results borrowed from somebody else’s project. We baseline each one on your data in the first fortnight, and report against that baseline monthly.
Turnaround time from application to decision
Measured on the whole queue rather than on the clean cases. Improvement has to come from removing rework, not from cherry-picking files.
First-time-right document submission
Share of files that pass document checks on the first attempt. This is where onboarding automation either shows up or does not.
Underwriter time per file
Preparation time separated from decision time, so the saving is visible where it happens and the decision is not rushed to make a number.
Roll-forward rate in early buckets
Accounts moving from the earliest delinquency bucket into the next one, alongside complaint volume — capacity gained at the cost of conduct is not a gain.
What we design around
The constraints come first, not at the security review.
Accountability stays with the regulated entity
RBI's expectations on outsourcing of information technology and on digital lending are clear that responsibility cannot be transferred to a vendor. We build to be inspectable, contractually and technically, and we scope engagements so your compliance team can evidence control.
Data residency and localisation
Payment system data has a storage-in-India requirement, and several institutions apply a stricter internal standard. We deploy in-country by default and treat any cross-border processing as an explicit, documented decision rather than an infrastructure default.
Explainability and human decisioning
Credit, pricing, rejection and collections treatment stay human decisions. Every automated step carries an immutable log of inputs, model version and output, so a decision can be reconstructed for an auditor months later.
Customer conduct and communication rules
Fair practice codes, grievance redressal timelines and commercial communication regulations set the boundaries for any automated contact. Calling windows, identification, escalation paths and opt-outs are built in, not configured later.
This is how we scope and build, not legal advice. Your compliance, risk and legal teams stay the authority on what applies to your organisation, and we work to their reading of it.
Your first project
Start narrow. Prove the number. Then scale it.
Take a single product's onboarding queue — a personal loan, an MSME working capital line, a motor policy — and put document intelligence in front of it for six weeks, running alongside the existing team rather than replacing it. You get a measured first-time-right rate on your own document mix, an exception taxonomy worth more than the automation itself, and an audit trail your risk function can review before anything scales.
Discovery Sprint
1–2 weeksWe sit with your operators, map the workflows, and score every candidate use case on value, data readiness and risk. You leave with a ranked backlog and a cost model.
- Use-case scorecard
- Data readiness report
- Costed roadmap
Proof of Value
3–4 weeksWe build the single highest-value agent against your real data and measure it against the humans doing that job today. If the number is not there, we say so.
- Working prototype
- Evaluation set
- Benchmark vs. baseline
Bring us one workflow from your floor.
Describe how BFSI works in your organisation — the systems, the constraints, the part that goes wrong every week. We will tell you whether it is worth automating and what the first project would cost.