AI agents for NBFC lending, on a model you own.
Vaman builds small language models trained on your loan data — collections, EMI, acquisition, onboarding, and policy — and runs them inside your cloud. Every action goes on the record.
Four problems. One platform.
These aren't technology problems. They're the operational patterns that limit every NBFC at scale.
Systems that don't talk
Your CRM has lead data. Your LOS has application status. Your LMS has payment history. Three systems, three teams, no shared view of a borrower's journey.
Borrowers who can't be reached
The borrower registered a number that's changed. The email bounces. The WhatsApp is on a different device. By the time a person tries a third channel, the borrower has moved on.
Single-channel collections
Call-only or SMS-only strategies lose borrowers who don't answer that channel. Without channel intelligence, contact rates stay low regardless of how many calls are made.
No end-to-end view
Nobody sees one borrower's path from application to disbursement to repayment. The collections team has no acquisition context. The servicing team has no collections history.
Governed agents across the lending journey.
Each agent has one job, reads your systems, acts within limits you set, and writes the outcome back.
Qualification Agent
Responds to new leads from web, WhatsApp and referral channels, asks your eligibility questions, scores each lead and routes qualified ones to a relationship manager.
Outbound Engagement Agent
Follows up with prospects by call and WhatsApp, answers questions from your approved content, and books callbacks with your team.
Application & KYC Agent
Guides applicants through the application, chases missing documents on WhatsApp, and submits complete files to your LOS.
Onboarding Agent
Sends disbursement confirmations, welcome messages and the EMI schedule, so borrowers know what to expect from day one.
EMI Reminder Agent
Reminds borrowers before each due date by voice, WhatsApp or SMS, shares your payment link, and confirms once your LMS records the payment.
Delinquency Prevention Agent
Spots accounts likely to slip to the next bucket and engages them early with options you have approved.
Collections Agent
Contacts overdue borrowers within permitted hours, offers repayment options from your policy, and logs every promise to pay.
Field Collections Copilot
Gives field officers each borrower's history, the next best action and a visit plan, and records the outcome.
Approved conduct policies, enforced in every call and message.
Vaman provides controls and evidence that help you meet your obligations. Your NBFC remains responsible for compliance.
A model trained on your portfolio, not everyone's.
Your Vaman model learns from your outcomes, your borrowers' languages, your products and your scripts. Before you commit, we benchmark it against frontier models on your own data for accuracy, cost and speed. The weights are yours.
Request a benchmark on your data →| What we measure | Vaman model | Frontier |
|---|---|---|
| Intent & response understanding | Measured on your data | Measured on your data |
| Correct repayment option offered | Measured on your data | Measured on your data |
| Cost per 1,000 conversations | Measured on your data | Measured on your data |
| Response speed | Measured on your data | Measured on your data |
Replaces nothing. Connects to what you already run.
Vaman reads your CRM, LOS, LMS, contact centre and collections systems, acts on borrowers, and writes every outcome back. Your systems stay the record. We don't replace anything — we add a governed intelligence layer on top.
Your cloud, in India, under your audit.
Run Vaman in your own cloud account in an India region or on your premises. Your contract includes audit and inspection rights, an exit plan and platform code escrow.
Visit the Trust centre →Live on one use case within a quarter.
No lengthy RFP cycles. No 18-month implementations. We start with one use case, prove measurable impact, then expand.
Readiness assessment
Three weeks · priority use cases, data readiness, conduct gap map, pilot plan.
Pilot on base models
Run on Vaman base models — fast to live, real results from week one.
Your trained model
Your own model in production, trained on your portfolio data.
More lending stages
Expand across collections, servicing, acquisition — same platform.
Your first 90 days, measured.
No synthetic benchmarks. Measured on your portfolio, in your environment, from day one of the pilot.
Collections
- —Contact rate by DPD bucket
- —Promise-to-pay kept rate
- —Roll-rate between DPD buckets
- —Complaints per 10,000 contacts
Servicing
- —On-time EMI rate
- —Payment confirmation time
- —Servicing queries resolved without escalation
Acquisition
- —Lead response time
- —Lead-to-application rate
- —Application completion rate
Answers for your risk committee.
Does the AI make credit or settlement decisions?
No. Agents gather information, remind and engage. People decide. Settlement offers require pre-approval from your credit and collections policy team. The agent selects from that approved menu — it does not negotiate outside those bounds.
Where does borrower data live?
In your cloud account in an India region, or on your premises. It is never used to train another client's model. Your data is yours, and the contract says so.
Who owns the model?
You do, including the weights. When you leave, the model and its training artefacts leave with you. The exit plan and platform code escrow are in every contract.
Can we stop it instantly?
Yes. There is a kill switch per agent, per campaign and for everything — available to your operations team at any time, without needing to call us.
How do you handle calling hours and conduct rules?
They are enforced by the platform before any call or message goes out, and every action is logged for review. A call scheduled for 7:40 pm is held and rescheduled to 8:00 am — automatically.
What if we stop working with Vaman?
Your model and data stay with you. The exit plan and platform escrow are in the contract. We help you run the model on your own infrastructure if you choose to.
Start with three weeks.
The NBFC readiness assessment gives you priority use cases, a data-readiness view, a conduct and DPDP gap map, and a pilot plan.
Book an NBFC readiness assessment