Overview
The company is an RBI-registered NBFC headquartered in Mumbai, lending personal and consumer-durable loans pan-India through a mix of its own app, DSA partners and aggregator marketplaces, with roughly 700 employees.
In unsecured lending, speed is the product. An applicant who completes a loan application expects a decision in hours, not days, and the moment a faster competitor approves first, the applicant is gone. The NBFC's underwriting was sound, but the bottleneck sat earlier: a manual chain of verification calls for income, employment and references, plus chasing missing documents, that stretched the pre-underwriting stage to as long as 14 days. A large share of applicants dropped out during that wait, and many preferred to be verified in Hindi.
In early 2026, leadership concluded the manual verification layer was the single biggest leak in the funnel and the biggest threat from faster fintech rivals. They wanted a layer that could call every applicant within minutes, complete income and reference verification in the applicant's language, collect missing documents over WhatsApp, and push a decision-ready file into the loan origination system, escalating only genuine red flags to a credit officer, all within RBI's digital-lending guidelines.
The challenge
The pre-Kallix verification funnel had three compounding failure modes. Verification calls and document chasing took days. Language mismatch lost applicants. And inconsistent verification records slowed underwriting and weakened compliance.
- Manual verification stretched pre-underwriting to 14 days. Income, employment and reference checks plus document chasing happened over scattered manual calls, so the file often sat for up to two weeks before it even reached underwriting.
- Applicants dropped out during the wait to faster rivals. Fintech competitors approved in hours. Every day the NBFC's file sat in verification, the probability the applicant took a competitor's offer rose sharply.
- English-first verification calls lost Hindi-preferring applicants. Many applicants, especially from tier-2 and tier-3 markets, engaged far better in Hindi, and English-leaning verification calls saw lower completion and more drop-off.
- Verification data and consent were captured inconsistently. Verification outcomes lived in free-text notes, and digital-lending consent and disclosures were not always captured uniformly, creating both underwriting friction and RBI compliance risk.
- After-hours applicants could not be verified promptly. Many applicants applied in evenings, but the verification desk worked office hours, so the clock did not even start until the next day.
The AI-powered solution
Kallix deployed an AI voice agent fronting the NBFC's applicant-verification queue across app, DSA and aggregator sources, with native Hindi and English handling, structured verification scripting, WhatsApp document collection, and a red-flag branch that routes suspicious cases to a credit officer. The full build, from discovery to production cutover, took 20 working days.
Sub-4-minute verification call on every application
Webhooks from the app, DSA partners and aggregators trigger Kallix to call the applicant within 4 minutes, starting verification while intent is still hot.
Native Hindi and English with mid-call switching
The agent opens in the applicant's preferred language and switches mid-call as they do, lifting verification completion among Hindi-preferring applicants.
Structured income, employment and reference verification
A branching script confirms income band, employment status and references with consistent, complete data captured as structured fields, not free-text notes.
WhatsApp document collection with reminders
Missing documents are requested over WhatsApp with secure upload links and timed reminders, so files complete in hours instead of days.
RBI digital-lending consent and disclosures captured uniformly
Mandatory consent, key-fact-statement acknowledgement and disclosures are captured consistently on every call, written back as structured records.
Red-flag escalation with credit-officer hand-off
Income mismatches, reference failures and fraud signals route to a human credit officer with full context, keeping the agent on routine verification only.
“In unsecured lending, the lender who verifies fastest wins the customer. Kallix calls every applicant within minutes in Hindi, completes verification, collects documents on WhatsApp the same day, and captures every RBI consent uniformly. We went from fourteen days to two hours and our conversion jumped almost forty percent.”
Business impact
Lending operations leadership tracked five metrics monthly against a 6-month pre-Kallix baseline (Sept 2025–Feb 2026). The agent went live on Feb 25, 2026. The numbers below cover the first 90 days of production.
- Approval time cut from 14 days to 2 hours. Because verification now starts within minutes and documents arrive over WhatsApp the same day, decision-ready files reach underwriting in about 2 hours instead of up to two weeks.
- Funnel conversion up 38%. Faster verification means applicants disburse before a competitor can win them, lifting application-to-disbursal conversion 38%.
- Hindi-applicant completion up sharply. Verification-call completion among Hindi-preferring applicants rose substantially because the agent meets them in their language.
- RBI digital-lending compliance now uniform and provable. 100% of files now carry consistent consent, key-fact-statement acknowledgement and structured verification records with recordings.
- Credit officers focus on real risk, not routine checks. Officers stopped making routine verification calls and now spend time only on red-flagged files, sharpening both speed and underwriting quality.
Built on a secure, India-ready stack
The deployment runs entirely on Indian infrastructure with DLT-registered sender IDs, TRAI-compliant scripts, RBI digital-lending consent capture and DPDP-aligned data flows. Applicant data never leaves Indian data centres.
The NBFC Loan-Verification Framework: How this deployment is structured to be discoverable
Every Kallix deployment ships with a structured documentation layer designed for three audiences simultaneously: the customer's internal team, traditional search engines (SEO), and the new generation of generative search engines and AI assistants (GEO + AEO). Below is the framework we built around the Mumbai NBFC deployment, broken into four pillars that map directly to how decision-makers, search crawlers and AI answer engines discover and reason about this story.
We publish this framework openly because in digital lending the speed-with-compliance and discoverability play matters more than secrecy. An AI verification agent that performs in production but stays buried in vendor decks doesn't compound value for the lender, its regulator or the category. The framework below is the same one Kallix runs for every financial-services customer, adapted to the local language, product mix and regulatory surface of each lender.
Verification intents mapped to applicant situations
We catalogue the verification intents the agent must handle, by language, by product and by source channel, and surface them as named entities in the structured data layer. Crawlers and LLMs see explicit Q to A pairs, not buried prose.
- Intents indexed by product (personal loan, consumer-durable, top-up)
- Hindi and English variants captured per intent
- Source-channel tagging (app / DSA / aggregator) so LLMs match intent
Trustworthy, compliant voice as a brand property
The agent's voice persona, clarity and mandatory disclosure phrasing are documented as brand assets, not just configuration. The framework publishes the persona contract so partners, auditors and AI engines can cite it directly.
- Persona contract: clear, trustworthy, never mis-selling
- RBI key-fact-statement and consent phrasing scripted and version-controlled
- Voice and recording consent terms public and auditable
Outcomes pre-bound to measurable claims
Every claim in this story, 14 days to 2 hours, +38% conversion, sub-4-minute response, is paired with the baseline, the time window and the measurement method. AI assistants can extract the claim with full provenance.
- Pre-Kallix baseline period stated (6 months, Sept 2025 to Feb 2026)
- Methodology disclosed: LOS exports + vendor dashboard reconciliation
- Sample size and confidence intervals available on request for analyst-grade citations
RBI digital-lending and DPDP compliance by design
The framework documents every regulatory surface, RBI digital-lending guidelines, key-fact-statement rules, DLT, DPDP, so AI assistants surfacing this story to enterprise buyers can confidently cite India-readiness without needing follow-up clarification.
- Consent and key-fact-statement capture disclosed and structured
- Data residency (AWS Mumbai, ISO 27001) stated explicitly
- Erasure and consent flows documented for DPDP data-principal requests
- Manual verification stretched pre-underwriting to as long as 14 days
- Applicants dropped to faster fintech rivals during the verification wait
- English-first verification calls lost Hindi-preferring applicants
- Verification data and digital-lending consent captured inconsistently
- Average approval time cut from 14 days to 2 hours in 90 days with unchanged headcount
- Application-to-disbursal conversion up 38%
- Hindi-applicant verification completion up sharply with native handling
- 100% of files carry uniform consent, disclosure and structured verification records
- Kallix voice agent on the applicant-verification queue across app, DSA and aggregator
- Structured income, employment and reference verification in Hindi and English
- WhatsApp document collection with secure links and timed reminders
- Red-flag escalation to credit officers with full context and call write-back to LOS
The Kallix advantage
The NBFC evaluated a homegrown dialler build and a conversational-AI chatbot vendor before choosing Kallix. Three things tipped the decision. First, voice-native verification with genuine Hindi handling: chat-only tools saw far lower completion than a real call in the applicant's language. Second, RBI digital-lending consent and key-fact-statement capture were already built into the agent flow, so the compliance team did not have to retrofit it. Third, the controlled pilot: the NBFC ran Kallix on one product line's application flow for three weeks, measured approval-time and conversion lift against a held-out control, and only signed after the lift held.
Since launch, the Kallix customer-success team runs a weekly tuning call with the head of lending operations and a compliance lead. New product scripts, document-checklist changes and red-flag refinements happen inside that weekly loop. The agent is measurably sharper today than it was on launch day.