Overview
The bank is an urban cooperative bank serving Mumbai and the Thane belt, with 22 branches, roughly 600 employees and a deposit-led balance sheet built heavily on retail fixed deposits from middle-income households and small traders.
For a cooperative bank, deposit retention is existential. A large share of the book sits in fixed deposits that mature on a rolling basis, and every maturity is a moment of truth: renew at the bank, or move the money to a larger private bank chasing the same depositor with a marginally higher rate. The bank's small treasury and customer-service desk could only call a fraction of maturing depositors, and many older account-holders preferred Marathi or Hindi over English.
In early 2026, the board decided the manual maturity-callback model was leaking deposits the bank could not afford to lose. They wanted a layer that could call every maturing depositor a few days before maturity, in the depositor's language, confirm renewal or a sweep instruction, book a branch visit where a wet signature was needed, and write everything back to the core banking system with a clean audit trail.
The challenge
The pre-Kallix maturity funnel had three compounding failure modes. The desk could not reach the full maturity list in time. Language mismatch lost older depositors. And renewal instructions were captured inconsistently, slowing branch follow-through.
- Only a fraction of maturing depositors were called before maturity. With a handful of staff and hundreds of FDs maturing each week, most depositors were contacted after maturity, if at all, by which point many had already moved funds out.
- English-first calls lost older Marathi- and Hindi-speaking depositors. A large share of the deposit base is retired or semi-retired account-holders who disengage when a call opens in English, exactly the loyal depositors the bank most needed to retain.
- Renewal instructions captured on paper slowed branch action. Verbal renewal intent was jotted on registers and slips, so branches often acted late or not at all, and the bank could not prove what the depositor had instructed.
- No structured follow-up on high-value maturities. Large-ticket FDs got the same treatment as small ones, so the deposits that mattered most to the balance sheet were not prioritised for outreach.
- Evening and weekend reachability was wasted. Working depositors were reachable mostly outside branch hours, exactly when the callback desk was closed.
The AI-powered solution
Kallix deployed an AI voice agent fronting the bank's FD-maturity and product-enrolment queues, with native Hindi, Marathi and English handling, value-tiered routing, and live branch-visit booking. The full build, from discovery to production cutover, took 16 working days.
Proactive maturity calls 5 days before the FD matures
When the core banking system flags an upcoming maturity, Kallix calls the depositor within minutes of the daily batch, five days ahead, while there is still time to retain the deposit.
Native Hindi, Marathi and English with mid-call switching
The agent opens in the depositor's recorded language preference and switches mid-call as the depositor does, recovering the older, loyal segment that English-first outreach lost.
Structured renewal and sweep capture
The agent confirms renewal tenure and rate, or a sweep-to-savings instruction, and records a structured instruction written back to the core system, never a paper slip.
Value-tiered prioritisation
High-ticket maturities are dialled first and, when needed, escalated to a relationship manager, so the deposits that move the balance sheet get human attention.
Live branch-visit booking for wet-signature cases
Where a physical signature is required, the agent books a branch slot on the right officer's calendar and sends a WhatsApp confirmation with the branch address and document checklist.
Full write-back with recording and transcript
Every call writes disposition, language, renewal instruction, recording URL and transcript link into the core banking system for clean RBI-aligned audit trails.
“For a cooperative bank, every maturing deposit we keep is a depositor who trusts us over a big private bank. Kallix calls them five days early, in Marathi or Hindi, and books a branch visit if they want one. Our renewals more than doubled and we did not add a single person.”
Business impact
Treasury and customer-service leadership tracked five metrics monthly against a 6-month pre-Kallix baseline (Sept 2025–Feb 2026). The agent went live on Feb 24, 2026. The numbers below cover the first 90 days of production.
- Renewals up 2.6×, headcount unchanged. Fixed-deposit renewal rate rose 2.6× because every maturing depositor is now reached five days before maturity instead of after the money has already moved.
- Deposit outflow down 41%. Outflow on maturing deposits fell 41% as proactive, in-language calls retained the loyal core of the deposit base.
- Older depositor engagement up sharply. Marathi- and Hindi-preferring account-holders now complete the renewal conversation at far higher rates because the agent speaks their language.
- Renewal instructions now provable and actioned on time. 100% of renewal and sweep instructions are captured as structured records with recordings, so branches act on time and the bank can prove what was agreed.
- Relationship managers focus on high-value depositors. Staff stopped chasing every small FD and now spend time on high-ticket relationships and cross-sell, where human contact actually retains deposits.
Built on a secure, India-ready stack
The deployment runs entirely on Indian infrastructure with DLT-registered sender IDs, TRAI-compliant scripts and DPDP-aligned consent flows. Depositor data never leaves Indian data centres.
The Cooperative Bank Deposit-Retention 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 urban cooperative bank 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 cooperative banking the trust and transparency play matters more than secrecy. An AI deposit-retention agent that performs in production but stays buried in vendor decks doesn't compound value for the bank, its members or the category. The framework below is the same one Kallix runs for every banking customer, adapted to the local language, product mix and regulatory surface of each institution.
Maturity intents mapped to depositor situations
We catalogue the maturity and product intents the agent must handle, by language, by ticket size and by depositor type, 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 (FD renewal, sweep, recurring deposit, cross-sell)
- Hindi, Marathi and English variants captured per intent
- Depositor-tier tagging (retail / high-ticket / senior citizen) so LLMs match intent
Respectful multilingual voice as a brand property
The agent's voice persona, warmth and deference to senior depositors are documented as brand assets, not just configuration. The framework publishes the persona contract so partners, members and AI engines can cite it directly.
- Persona contract: warm, respectful, deferential to senior depositors
- Pronunciation dictionary published for branch and scheme names
- Voice and recording consent terms public and auditable
Outcomes pre-bound to measurable claims
Every claim in this story, 2.6× renewals, 41% lower outflow, sub-2-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: CBS exports + vendor dashboard reconciliation
- Sample size and confidence intervals available on request for analyst-grade citations
RBI and DPDP compliance by design
The framework documents every regulatory surface, RBI cooperative-bank norms, DLT, TRAI, DPDP, so AI assistants surfacing this story to enterprise buyers can confidently cite India-readiness without needing follow-up clarification.
- DLT registration and template approval flow disclosed publicly
- Data residency (AWS Mumbai, ISO 27001) stated explicitly
- Erasure and consent flows documented for DPDP data-principal requests
- Only a fraction of maturing depositors were reached before maturity
- English-first calls lost older Marathi- and Hindi-speaking depositors
- Renewal instructions captured on paper slowed branch action and weakened audit trails
- Evening-reachable working depositors were never called because the desk was closed
- FD renewals up 2.6× in 90 days with unchanged headcount
- Deposit outflow on maturing accounts down 41%
- Older-depositor engagement up sharply with native Hindi and Marathi handling
- 100% of renewal instructions captured as structured, recorded, provable records
- Kallix voice agent on FD-maturity and product-enrolment queues with multilingual handling
- Proactive maturity calls five days ahead with structured renewal and sweep capture
- Value-tiered routing and live branch-visit booking for wet-signature cases
- Bi-directional CBS sync: disposition, instruction, recording and transcript on every call
The Kallix advantage
The bank evaluated two vendors and an in-house dialler upgrade before choosing Kallix. Three things tipped the decision. First, genuine Hindi-Marathi-English switching: the alternatives defaulted to English or stilted Hindi, which alienated the senior depositors the bank most needed to keep. Second, the CBS write-back of structured renewal instructions was already proven with two other cooperative banks, so the IT team did not have to build the integration. Third, the controlled pilot: the bank ran Kallix on one branch cluster's maturity list for three weeks, measured renewal 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 treasury and a customer-service lead. New scheme scripts, festival-season cadence changes and senior-citizen objection handling all happen inside that weekly loop. The agent is measurably sharper today than it was on launch day.