Case studies, measured problem by problem
Each engagement starts with a matrix: the problem, what we built, the product behind it and the metric we agreed to move.
Antara by Max Group Healthcare and senior care · In production
Antara runs customer experience, e-commerce personalisation, an AI concierge for seniors and its EHR and facility operations on PineOS.
| Problem | What we built | Product | Metric | Result |
|---|---|---|---|---|
| Family enquiries arrive by phone, web and WhatsApp with no shared view | Voice and chat concierge that profiles callers and books visits and services | Nexus | Share of enquiries resolved by AI | 62% |
| Resident records kept on paper and spreadsheets | Antara EHR: onboarding, assessments, prescriptions and care plans | CareOS | Time to onboard a resident | 3 days to 6 hrs |
| Facility and service requests lost between teams | Smart ticketing with routing and SLA tracking | CareOS | Median ticket closure time | 48% faster |
| Generic product suggestions in the online store | Personalised search and recommendations | Nexus | Conversion on recommended items | +22% |
Motilal Oswal Financial services · In production
Motilal Oswal uses PineOS for customer support, data AI and its loyalty program.
| Problem | What we built | Product | Metric | Result |
|---|---|---|---|---|
| High volume of repetitive support queries | Support agents on chat and voice with escalation to human advisors | Nexus | First-contact resolution rate | +19 pts |
| Business questions wait days for reports | Gen BI over governed customer and transaction data | DataOS | Time to answer a business question | 2 days to 5 min |
| Low repeat engagement from existing customers | AI-driven loyalty program with personalised offers | Nexus | Repeat engagement rate | +27% |
Public sector bank Public sector banking · Proofs of concept
Three proofs of concept on banking agents: vehicle loan digitisation, a revenue leakage algorithm and an AI concierge for product adoption.
| Problem | What we built | Product | Metric | Result |
|---|---|---|---|---|
| Vehicle loan applications processed by hand | Document capture, KYC checks and decision support across the loan journey | BeCom + Studio | Loan turnaround time | 5 days to 1 day |
| Fees and charges missed or wrongly waived | Revenue leakage algorithm across transactions and accounts | DataOS | Leakage identified and recovered | ₹4.2 Cr |
| Customers unaware of relevant banking products | AI concierge that recommends products in the customer language | Nexus | Product adoption rate | +14% |
| KYC documents checked manually | OCR capture with rule checks and a reviewer queue | BeCom | KYC cases cleared per day | 3.5x |
Mining major Mining · In production
One of India’s largest state-owned miners uses PineOS for tender management, dumper offloading, ore quality checks and maintenance of excavation equipment.
| Problem | What we built | Product | Metric | Result |
|---|---|---|---|---|
| Tender documents reviewed manually | Tender agent that extracts requirements and compares bids | Studio | Tender evaluation time | 60% faster |
| Ore quality checked by sampling | Vision-language model grading ore from camera feeds | Studio | Share of loads graded automatically | 85% |
| Queues at dumper offloading points | Offloading scheduling and live tracking | Quantm | Dumper idle time | -23% |
| Unplanned failures of excavators, pulleys and ropes | Visual inspection with predictive maintenance alerts | Studio + DataOS | Unplanned downtime | -31% |
Battery manufacturer Manufacturing and energy · Proof of concept
A proof of concept covering ESG report automation and agents that test business workflows.
| Problem | What we built | Product | Metric | Result |
|---|---|---|---|---|
| ESG data scattered across plant reports | Document AI that extracts ESG metrics into reporting templates | BeCom | Hours per ESG report | 120 to 30 hrs |
| Workflow changes tested by hand | Agents that run end-to-end workflow tests | Studio | Test cycle time | 55% faster |