Business Performance Highlights
AUM Growth:
- Retail AUM crossed ₹1 lakh crore mark
- September 2026 Total AUM increased 24% Year-over-Year (YoY)
- September 2026 Growth AUM increased 30% YoY
- March 2026 AUM growth of 25% YoY
Financial Guidance:
- FY27 total PAT guidance of ₹2,200-2,300 crore
- Guidance primarily led by Growth business PBT (excluding POCI recovery related profits)
- Working towards 20-25% AUM CAGR along with Return on Equity (RoE) of 15%+
AI Implementation and Impact
Productivity Improvements:
- Improved productivity by ~80%
- Monthly disbursement per employee: ₹5.1 lakh (Sep-26) vs ₹4.1 lakh (Sep-25) - 33% YoY growth with AI driving half the uplift
- AUM per employee: ₹27.5 lakh (Sep-26) vs ₹20.7 lakh (Sep-25) - 23% YoY growth with AI driving half the uplift
Risk Reduction:
- Reduced credit risk by 50%
New Business Creation:
- Created new lines of businesses that disburse ₹4,000 crore every quarter
AI Infrastructure Scale:
- Tokens processed increased from 23 billion to 476 billion per quarter (20x increase)
- Cost per token down 46%
- ~₹6 crore API spend avoided in 5 quarters (FY26-Q2 to FY27-Q2 volumes priced at FY25-Q3 unit cost)
AI Technology Architecture
Hybrid Approach:
- Frontier APIs (Anthropic, Google, OpenAI) handle 36% of tokens for complex reasoning and new use cases
- In-house models (fine-tuned open-source) handle 64% of tokens for high-volume, repeatable tasks
- Proprietary models trained on contextual data
GPU Infrastructure:
- In-house GPU cluster with NVIDIA H100/RTX PRO 6000 for model training and fine-tuning
- Production deployment runs on cloud for scalable high availability and low latency
Benefits:
- Lower cost per token
- Data stays in-house
- No single-vendor lock-in
- Faster, tuned responses
Research and Intellectual Property
Conference Publications:
- "Toxicity Detection in Romanized Indic Languages" - ICON 2026 (2,28,198 messages in 8 Indian languages, 81.4% accuracy)
- "Semantic Density Does Not Predict Agent-Memory Collapse" - NeurIPS 2026 PALM Workshop, Paris
Patent Activity:
- Patenting ideas and sharing them worldwide
Target Market Focus
Bharat - Middle India:
- Focus on serving Kirana owners, Tailors, Small transporters, Textiles owners, Small dairy operators
- Not targeting Affluent, Urban customers
- Rural business in early days
Addressing Challenges:
- Thin documents with incomplete & single names
- Weak 1/2 line addresses
- Generic addresses without flat names or street names
AI Agents and Applications
Core Lending Functions:
- Multiple agents helping across functions
- Star agents shape core lending functions from Sales to Collections
- AI now part of entire lending process
Company Representatives
- Bipin Singh: Company Secretary (signed the regulatory filing)
- Ruchika Jain: VP-IR and Sustainability
- Ashish Raj: Senior Manager - IR and Strategy
Website and Investor Relations
- Presentation available on company website: www.piramalfinance.com
- Investor contact: investor.relations@piramal.com