Your data. Your model.
Your intelligence. Your control.
Vaman builds AI models Indian financial institutions own, runs them inside your cloud, and puts every action on the record. Start with one focused workflow, govern every action with verifiable policy gates, and scale safely across the institution.
Three-week readiness assessment · Fixed fee · Deployed in your environment
Renting AI got you a pilot.
Owning it gets you to production.
Financial institutions have tried two routes to AI. Both stall before production.
Fast to start.
But your data leaves the building. The intelligence you build isn't yours. The model can change under you without notice. Your pilot stays a pilot.
Full control.
But it is slow and costly. Governance is rebuilt from scratch for every project. A small team maintains it forever. Most institutions never reach production.
Control, without starting from zero.
We build a model on your data. You own the weights. Our platform governs every action from day one. The intelligence you build stays yours, in your cloud.
Small models, trained on your work, beat giants at it.
A general model knows a little about everything. Your model knows your borrowers, your products and your rules. We prove the difference before you commit: every pilot includes a head-to-head benchmark run on your own data, with agreed test cases before you sign off.
Where a frontier model performs better on your tasks, we use it — inside your cloud.
Request a benchmark on your data| Metric | Your Vaman model | Frontier model |
|---|---|---|
| Accuracy on your tasks | 94% on your workflows | 71% on your workflows |
| Cost per 1,000 interactions | ~₹18 | ~₹220 |
| Response speed | < 800 ms | < 1,200 ms |
| Data leaves your cloud? | Never | Always |
Indicative benchmarks run on NBFC collections data. Your numbers will vary — we prove it on your data before you commit.
We build the model. You own it.
Our platform keeps it governed.
- Trained model weights — yours to keep if you ever leave
- Every AI output and audit record, stored in your cloud
- Inspect, port or exit at any time — no lock-in
- Route (model selector), Studio (agent builder), Ledger (audit log)
- Watch (quality monitoring) and Guard (policy enforcement)
- Platform improves every quarter — you benefit automatically
- End-to-end model training, fine-tuning and MLOps
- Security benchmarking and regulatory readiness checks
- You pay your cloud provider directly — no markup on compute
From first lead to final recovery.
Governed AI agents engage borrowers on voice, WhatsApp, SMS and email, read your CRM, LOS and LMS, and write every outcome back.
Respond to every lead fast, and hand qualified ones to your team.
Guide borrowers through documents and KYC without chasing by hand.
Remind before the due date, on the channel each borrower actually answers.
Engage early, inside RBI’s conduct rules, and route sensitive cases to people.
Every action on the record.
For every message, call or recommendation your agents make, you can see the data, model version and person behind it.
Your data stays in India.
Or inside your walls.
Run Vaman in your own cloud account in an India region, or on your premises. Your data never trains another client's model, and you and your auditors can inspect everything.
Your cloud account, India region
Run Vaman in your own AWS, Azure or GCP account in an India region. We manage the deployment under your access controls.
On-premise
Deploy inside your own data centre. Your team runs the infrastructure; we provide support and platform updates.
Vaman-hosted in India
Hosted in Vaman's cloud account in an India region. Ask us for current availability and the contract terms.
From first conversation to production in four steps.
Readiness assessment
Three weeks, fixed fee. Priority use cases, data readiness, regulatory gaps and a pilot plan your board can approve.
Pilot
One use case live on Vaman base models, judged against a scorecard agreed before we start.
Production
Your own trained model, benchmarked against frontier models, governed and running in your environment.
Scale
More use cases on the same platform, with no new governance build and no new vendor agreement.
What we measure in your first 90 days.
We publish results only when they are measured and our clients agree. These are the measures every pilot is judged on.
Contact rate
Measured in your pilot
Promise-to-pay kept
Measured in your pilot
Roll-rate between DPD buckets
Measured in your pilot
On-time EMI rate
Measured in your pilot
Lead response time
Measured in your pilot
Complaints per 10,000 contacts
Measured in your pilot
Regulation, decoded.
RBI FREE-AI: what NBFCs need to do before the next Monetary Policy
The Reserve Bank's Fintech Regulatory Engagement (FREE-AI) framework sets expectations for algorithmic decision-making in regulated entities. Here is what the circular means for NBFC lending operations.
Read →Keeping AI collections inside RBI's recovery conduct rules
Recovery Agent Guidelines, the Fair Practices Code and the Digital Lending norms together set the guardrails for AI-assisted collections. How to configure automated engagement that stays inside all three.
Read →DPDP from May 2027: what lenders must fix before the compliance deadline
The Digital Personal Data Protection Act takes effect for regulated financial services in May 2027. The key obligations, what counts as a "significant data fiduciary", and where AI systems need to change.
Read →Own your AI. Start with three weeks.
A readiness assessment gives you a prioritised use-case plan, a data readiness report and a pilot scope your board can approve — in three weeks, for a fixed fee.