How Finance Buddha automated Hinglish lead qualification across SME, BL, LAP, OD and Gold loans with Caller Digital's Anjali
Finance Buddha (Finbud Financial Services Limited) operates one of India's most active personal- and SME-loan marketplaces. Speed-to-lead is the single biggest determinant of conversion — borrowers comparing offers across Finance Buddha, banks, NBFCs and competing marketplaces decide within hours of submitting their enquiry. Caller Digital deployed a Hinglish-first outbound voice agent (Anjali) plus an integrated WhatsApp engagement layer to call every lead in minutes, qualify them on 12 structured fields, and warm-transfer the highest-intent leads to senior counsellors with a clean pre-call brief.
Sub-15-minute speed-to-lead across SME, business and gold-loan enquiries — in Hinglish, with structured qualification — beyond what a manual SDR floor could maintain
Finance Buddha's counselling team needed to call every inbound borrower lead within minutes of capture and qualify them across loan amount, purpose, urgency, business vintage, existing lender offers and EMI burden — in Hindi, English and Hinglish — before warm-transferring the highest-intent leads to senior counsellors.
Manual SDR calling could not maintain sub-15-minute speed-to-lead at scale. Qualification fields were captured inconsistently across counsellors. Competitor offers — the single largest cause of leakage — were often discovered too late to defend against. And the lead-operations team had no clean way to brief senior counsellors with a structured pre-call summary, so seniors spent five-to-ten minutes reading lead history before every hand-off conversation.
Speed-to-lead slipping past the golden hour
Inbound borrower enquiries routed through digital and DSA channels often waited hours for a first call, by which point comparable offers from competing lenders had already landed.
Inconsistent BANT-style qualification
Different SDRs captured different field sets — sometimes loan amount only, sometimes business vintage only, rarely the existing-lender competitive context.
Competitor offers discovered too late
Borrowers comparing lenders typically only mentioned competitor names when explicitly probed. Without structured probing, that signal was lost.
Senior counsellor pre-call brief was manual
Hand-offs to senior counsellors required five-to-ten minutes of manual lead review per conversation — pure overhead burned before the closing pitch started.
Anjali — a custom-trained Hinglish-first outbound voice AI agent with stage-driven discovery, AI-inferred briefing fields, and WhatsApp fallback after unanswered calls
Caller Digital deployed Anjali — a custom-trained outbound voice agent introduced as a Finance Buddha relationship manager. Anjali runs a stage machine: greeting and name verification, value intro citing the pre-qualified offer, busy-nudge fallback (offer to drop details on WhatsApp), structured need discovery one question at a time (amount → purpose → urgency → business ownership → vintage → existing lender offers), customised pitch, application-link consent, and optional warm-transfer for QUALIFIED_HOT leads.
Every call ends with one of 15 structured dispositions — INTERESTED, CALLBACK_REQUESTED, NOT_INTERESTED, NOT_INTERESTED_RATE, NOT_INTERESTED_COMPETITOR, QUALIFIED_HOT, ESCALATED_TO_AGENT, INVALID_NUMBER, DND_REQUESTED, LANGUAGE_BARRIER, FOLLOW_UP_REQUIRED, NO_ANSWER, VOICEMAIL, BUSY, ALREADY_PROCESSED. Beyond the disposition, the agent captures 12 core qualification fields plus seven AI-inferred briefing fields (objection type, competitor name and rate quoted, sentiment, two-three-line AI summary, system-suggested next best action) surfaced for the senior counsellor's pre-call brief.
WhatsApp closes the loop. After 3–4 unanswered or rejected attempts, an approved WhatsApp template fires via the Aisensy Business API with the borrower's application link. Post-call follow-ups send automatically on INTERESTED or CALLBACK_REQUESTED dispositions; callback confirmations carry the agreed date and time; re-engagement nudges fire for leads marked FOLLOW_UP_REQUIRED.
Hinglish-first conversation, telephony-trained
Default register is Hinglish — natural code-mix the way Indian telecallers actually speak — with silent mirroring into Hindi or English based on the borrower's preference. Trained on real Indian telephony audio for SME and personal-loan lead profiles.
One-question-at-a-time discipline
Non-negotiable: the agent never stacks two questions in one utterance. Pacing is natural, with ~500ms pauses after greetings and before asking questions. Sentences kept short and breakable.
12 qualification fields + 7 AI-inferred briefing fields
Intent, loan purpose (enum), urgency (enum), business ownership, vintage, existing lender offers, current EMI burden, preferred callback time, WhatsApp consent — captured structurally. Objection type, competitor name and rate, sentiment, AI summary, next best action — inferred for the senior brief.
13-day connectivity analysis baked into the dialler
Pre-launch Finance Buddha measured connectivity by hour. 12 PM–2 PM peaks at 38–39% pickup; 9–10 AM dips to 22%. The dialler is configured to prioritise peak windows and de-prioritise low-yield slots.
WhatsApp fallback via Aisensy Business API
After 3–4 unanswered call attempts, the agent fires an approved WhatsApp template with the application link. Post-call follow-ups, callback confirmations, and re-engagement nudges are automated per disposition.
RBI Fair Practices Code enforced at the platform layer
No aggressive sales language. No commitment to specific interest rates or sanctioned amounts. Instant DND honour on opt-out keywords. Transparent disclosure of calling on behalf of Finance Buddha. Prohibited terms screened throughout.
Nine measurable performance standards configured for the POC, with the connectivity dialler and WhatsApp fallback delivering structured improvement against manual SDR baselines
Finance Buddha and Caller Digital signed the Service Agreement on 14 May 2026 with a 16-working-day deployment timeline. The POC is configured against nine measurable performance standards, measured monthly from 30 days post Go-Live.
| Dimension | Before | After |
|---|---|---|
| Speed-to-lead | Hours / next-day on manual SDR | Sub-15-minute outbound on every lead |
| Qualification field capture | Inconsistent across counsellors | 12 core + 7 AI-inferred fields per call |
| Competitor offer visibility | Discovered too late or not at all | Captured live with name + rate where shared |
| Senior counsellor pre-brief | 5–10 min manual review per hand-off | AI summary + next best action surfaced automatically |
| Unanswered-call workflow | Manual or skipped | WhatsApp template fires after 3–4 attempts via Aisensy |
| Language consistency | Varied by SDR shift | Hinglish-first on 100% of dispositioned calls |
Frequently Asked Questions
An outbound voice AI agent (Anjali) for first-level lead qualification across SME Loans, Business Loans, Loan Against Property, Overdrafts and Gold Loans, supported by a WhatsApp engagement layer for follow-ups and re-engagement after unanswered calls. The agent runs a stage-driven Hinglish-first conversation, captures 12 core qualification fields plus seven AI-inferred briefing fields, and warm-transfers the highest-intent leads to senior counsellors.
Hindi, English and Hinglish — the natural Hindi-English code-mix Indian telecallers use. The default register is Hinglish; the agent mirrors whichever language the borrower is most comfortable in. The voice persona (Anjali) is female, warm, professional, with a natural local Indian accent.
Caller Digital's agent enforces RBI FPC at the platform layer: no aggressive sales language, no commitment to specific interest rates or sanctioned amounts, immediate DND honour on opt-out keywords in English or Hindi, transparent disclosure that the agent is calling on behalf of Finance Buddha, and prohibited terms screening throughout the conversation. Format discipline is strict — output is conversational text only.
After 3–4 unanswered or rejected call attempts, the agent fires an approved WhatsApp template via the Aisensy WhatsApp Business API with the borrower's application link. Post-call follow-ups send automatically on INTERESTED or CALLBACK_REQUESTED dispositions; callback confirmations carry the agreed date and time; re-engagement nudges fire for leads marked FOLLOW_UP_REQUIRED.
Nine performance standards: ≥90% lead qualification accuracy, ≥95% information response accuracy, ≥95% lead information capture, ≥95% counsellor warm-transfer success, ≥90% escalation routing accuracy, ≥98% outbound dial-out success, ≥98% WhatsApp workflow execution, ≥99% retry logic execution, and ≤3-second P50 bot initial response latency. Measured monthly starting 30 days post Go-Live.
The POC is configured at ₹5/min on a 10,000-minute prepaid package with no enterprise minimums and no multi-year lock-in. Deployment ran 16 working days from Service Agreement signature (14 May 2026) through to UAT. Post-pilot pricing is reviewed jointly based on observed traffic and outcomes. Platform uptime is committed at 99.5% monthly with structured service credits below target.
Run a similar voice AI rollout for your team
We work with Indian enterprises across BFSI, D2C, retail SaaS, and consumer services. Book a 30-minute consult to see what voice AI can do for your call queue.
© 2025 Caller Digital | All Rights Reserved