Onemi Technology Solutions Ltd Q4 FY2026 Concall Summary

Key Takeaways

• AUM surged 73% YoY, quality improved in parallel: AUM reached ₹7,066 crore while GNPA fell 77 bps to 2.12%, a rare combination that management calls prudence as competitive advantage.
• Deliberate yield compression is the core strategy: The company is passing operating leverage and lower funding costs to customers to attract higher-quality borrowers, betting that lower credit costs will more than compensate over time.
• AI is not a slide deck, it is the operating backbone: Transformer models replaced gradient-boosted trees in underwriting, multi-agent systems hunt fraud, and AI voice agents now handle early-bucket collections at over 70% of human efficiency.
• LAP is the next growth vector, but years from maturity: The secured book reached ₹518 crore with 40% of customers sourced organically from the existing base, yet steady-state ROA remains a year or two away.
• What to watch next quarter: Whether the 40% AUM growth guidance holds without asset quality slippage, the pace of funding cost improvement from the A- rating, and any concrete timeline for the next credit rating upgrade.

The Deliberate Yield Pivot

• Strategy, not accident: The moderation in yields is a calibrated shift toward higher-quality customer segments who are priced lower but exhibit superior long-term credit behavior. Management calls this "a very deliberate strategic direction."
• Mechanics of the pass-through: As operating leverage, lower funding costs, and reduced risk materialize, the company transmits a large part of these benefits to customers, attracting better credit behavior and driving a steady decline in credit costs over years.
• Spread trajectory: Yield minus cost of borrowing sits at ~18–19% in the near term and will reduce over the longer term; average NIM spread over 3–4 quarters is 14–16%.
• Revenue growth lagged AUM growth: Revenue rose ~4–5% QoQ while portfolio yield rose ~1%, with the gap explained by small margin compression and a reduction in other income.
• The bet: Management believes the reduction in margin from higher-quality customers will be compensated by cost of fund improvement, operational leverage improving OPEX margin, and 10–15% improvement in credit cost over a sustained period.

Ranvir Singh: "The moderation and margins are substantially linked to a very calibrated shift towards high quality customer segments, which of course is priced lower, but they also exhibit a very superior long-term credit behavior. This is not incidental, it is a very deliberate strategic direction in which we are going."

Asset Quality: The Prudence Engine

• GNPA improvement was structural, not cosmetic: GNPA fell from 2.89% to 2.12% (77 bps YoY), Net NPA at 0.29%, and collection efficiency stayed above 97%. The improvement came from better borrower selection, tighter geography management, and disciplined collections.
• Early-warning systems drove real action: Disbursements were reduced by ~7% and lending paused across ~450 pin codes where early-warning flags detected elevated risk, part of a regular systematic exercise.
• Stage migration tells the story: Stage-2 book reduced from 3.5% to 2.4%; Stage-3 from 2.9% to 2.1%, reflecting portfolio movement driven by selection and collections discipline.
• New customers are less stretched: Customers acquired in FY26 carry FOIR of 30%, down from 34% the previous year, meaning lower financial burden relative to disposable income. First EMI T90 improved from 1.5% to 0.7%.
• Provisioning stayed conservative: Stage-3 ECL coverage at 86.2%, Stage-2 coverage improved from 59.2% to 75.6%, total PCR of 86%, and a management overlay of ₹136 crore held as buffer.
• The GNPA spike last year was a technical change, not credit deterioration: The write-off period was extended from 120 days to 150 days after discussion with auditors; the 150-day policy has been in place for two years and GNPA behavior has been consistent since.

Ranvir Singh: "The objective is not merely to identify stress after delinquency emerges - it is to detect behavioral deterioration before it surfaces in the number."

AI & Technology Moat

• Underwriting model leap: The core underwriting engine moved from gradient-boosted trees to transformer-based sequence models; AUC improved from the 60s to 74.
• Fraud detection is now multi-agent: A setup covering vision, text, tabular, and graph data, where agents surface new fraud hypotheses on their own when they spot anomalies.
• GenAI-augmented feature store: Monitors over 7,000 features for causal drift, catching behavioral deterioration before it shows up in delinquency numbers.
• Collections got the AI treatment: AI voice agents in early-bucket tele-collections operate at over 70% of human recovery efficiency; AI-assisted customer support cut response time from 200 seconds to under 90 seconds with 100% call-quality coverage.
• Internal productivity is AI-driven: More than 80% of internal code is now written with AI assistance. In LAP, AI-powered smart document processing improved FTR rates by over 30% and reduced credit queries by 50%.
• All core systems built in-house: Management positions the company as "a technology company that does lending, not the other way around."

Neha Shivran: "We have moved past the static rules to a multi-agent setup, covering vision, text, tabular, and graph data. And the agents are actually surfacing new fraud hypothesis on their own when they spot anomalies. In this environment, where fraud patterns mutate every week, that's a different kind of capability than just having a few more checks."

LAP: The Secured Pivot, Still Early

• Scale so far: LAP AUM at ₹518 crore across 98 branches in 8 states, representing 7.3% of total AUM, built over roughly two years.
• Organic demand from existing base: More than 40% of LAP customers come from the existing Kissht customer base, providing zero customer acquisition cost for that segment and helping branch economics.
• Conservative LTV stance: LTV at 48% and expected to remain range-bound around that level; the proposition is superior turnaround time and better offer, not higher LTV risk. Customers are tier-1 city, loans of ₹6–10 lakhs on property valued at ₹12–25 lakhs.
• Branch expansion planned: At least 80 more branches by end of FY27, with calibrated expansion in Karnataka, AP, Telangana, Tamil Nadu, UP, and Maharashtra.
• Profitability is distant: The LAP business has not reached breakeven and is still probably a year or two away from steady-state ROA; Q4 OPEX increase was partly driven by LAP branch costs.
• Will stay small relative to PL: LAP may continue to remain a small portion of total AUM for FY27 as personal loans grow at a decent rate.

Funding, Capital & Ratings

• Funding structure is balanced: 50% on-book (₹3,556 crore, supported by 45+ lending partners, cost of borrowing 11.5%–14.25%) and 50% off-book (₹3,510 crore, FLDG up to 5%).
• A- rating is already paying off: The February 2026 credit rating upgrade delivered a 200 bps improvement in marginal cost of borrowing, which management says is certain to sustain for the following year.
• Next upgrade in sight: Management is cautiously optimistic on getting another upgrade from Crisil and India Ratings at some point in FY27, with active discussion over the next 6 to 8 months; a further one-notch upgrade could bring 100–150 bps reduction in cost of borrowing, possibly in the later part of FY27 but more so in FY28.
• IPO proceeds deployed: 75% of the primary issue (~₹630 crore) already infused as fresh equity into the NBFC subsidiary Si Creva Capital; the remaining 25% earmarked for technology, AI infrastructure, brand building, and strategic objectives is being deployed in a calibrated manner.
• Capital adequacy comfortable: NBFC capital adequacy at 25.3%; net worth ₹1,343 crore (2.4x increase over 3 years through retained earnings and organic profitability). Management says the company is comfortably capitalized to achieve 40% AUM growth for FY27.
• Sustainable leverage: Steady-state on-book debt-to-equity of 2.5–3x, with potential to improve if the secured book becomes more meaningful, though that is a two-year journey. Off-balance sheet leverage faces no real constraint as only 5% FLDG capital is blocked.

Customer Franchise & Distribution

• Large base enables repeat-led growth: 1.18 crore unique customers with 0.35 crore active provides repeat-customer flexibility, ensuring a stable growth trajectory without over-reliance on new acquisition.
• Acquisition is diversified: Digital marketing contributes less than 50% of customer acquisition; other channels (merchants, e-commerce platforms, aggregators, organic) contribute ~55%, with organic share at 30% and growing.
• Cross-holdings reveal the positioning: About 45% of active customers have another personal loan outside OnEMI's, typically from a larger institution with a larger ticket size, suggesting OnEMI serves as an entry or supplementary lender.
• Brand trust accelerated: Association with Sachin Tendulkar has accelerated brand trust, particularly in mass-market and mass-affluent segments.
• Collections infrastructure is deep: Over 7,000 field agents and 1,000+ tele-callers, more than 95% collections in-house, covering 17,000+ pin codes with intervention from as early as 5 DPD. A dedicated NPA recovery team was created given projected 40%+ AUM growth.
• Product expansion on the horizon: Beyond personal loan and LAP, the consideration list over the next five years includes gold loan, business loan, education loan, plus fee-based cross-sell of insurance and mutual fund distribution through a newly created subsidiary. None are immediate.

Ranvir Singh: "First and foremost, a demonstrated credit performance, which is coming on the back of all the AI, ML models and all the things which I talked about on collections and that has really kept us in the good stead for the segment that we operate in. Secondly, given the large base of customer that we have, it does allow for a scalability, not only coming in form of repeat customers, but also in the form of you can further select within that high quality customers, which will ensure customers will not only give you scale, but also deliver a credit behavior which is far, far superior."

Guidance & Commitments

Key Metrics