The #1 Enterprise AI Growth Partner

Case Study • Insurance AI

Processing 10× More Claims at FGSPL with an End-to-End AI Workflow

Insurance

AI Consultancy + Development + Agents

June 2024

Throughput

10×

More Claims

OPEX Cut

83%

Cost Reduction

AI Claims Processing

Email Ingestion

Auto-Extract

Damage Vision

AI Assessment

Report Gen

Auto-Create

Live Processing50 Claims/Day

3min

Touch Time

Settlement

₹70

Cost/Claim

6h SLA

<0.5% Errors

AI Workflow

Executive Snapshot

Insurance Claims AI Transformation

Client

FGSPL – Third-party claims admin for a leading PSU insurer

Challenge

Manual, multi-step claim handling slowed throughput & drove costs

Services

AI ConsultancyAI DevelopmentAI Agents

Key Results

10× throughput

≈83% OPEX cut

<3min touch-time

6h settlement SLA

“This Solution let us handle a week’s workload before lunch.”

COO

FGSPL

Challenge & Objectives

Challenge Narrative

FGSPL’s adjusters juggled e-mails, voice notes, photos, garage calls and PSU templates to settle auto claims. Each file demanded data entry, image triage, API checks, report writing and SAP uploads—stretching processing time, stressing staff and delaying customer payouts.

KPI (Baseline → Target)

Speed to settle

72h → 6h

91% faster

Agent output / day

5 → 50

10× increase

OPEX / claim

₹420 → ₹70

83% reduction

Data-entry errors

6% → <0.5%

Eliminated

Solution Overview

End-to-End Claims AI Workflow

Service Pillars Engaged

AI Consultancy

3-week audit & SOP blueprint

AI Development

Full claim-bot pipeline, CRM adaptor

AI Agents

Orchestrating mail→report→approval

AI Workflow Process

Agent ingests PSU claim e-mails, auto-creates CRM leads

Chats with claimants via WhatsApp/SMS to gather docs

Runs vision model to grade damage & identify parts

Validates plates and licences via mParivahan

Compiles pre-filled loss report, routes single-click approval

Claims Management Dashboard

Live Claims Processing3628 Processed

New

Processing

Approved

201

Closed

Claim Settled

Jitendra Mandal

₹8,000

Approved

Chetan Thakur

₹12,500

Under Review

Neha Ahmed

₹15,000

Closed

Performance Metrics

93%

Touch Time Reduction

91%

SLA Improvement

Live Claims

Architecture & Approach

Implementation Strategy

1Discovery & SOP Design

14 ride-along sessions, 9 swim-lane maps

42 manual steps → 7 automated modules

2Deployment & Adoption

Pilot (2 adjusters) → full rollout (40) in 8 weeks

“Shadow-bot” mode for trust, then zero-touch

3Technical Build

Input ingestion: Mailgun API + OCR fallback

Damage vision: Detectron2 model, nightly reinforcement fine-tune

Govt verification: mParivahan & VAHAN APIs

Workflow orchestration: LangGraph + Kafka

Report composer: LaTeX → PDF, e-sign

Security: VPC-isolated, AES-256, RBAC, ELK audit logs

Claims Processing Flow

PSU claim e-mail

(PDF / JPG / voice)

Mail + OCR listener

Multi-channel comms bot

Vision model: damage & part detection

mParivahan / VAHAN API checks

Garage work-order

SAP / core-system update

Final AI report + closure pack

Automated Flow

Outcomes & Business Impact

Metric	Pre-9AI	Post-9AI	Δ
Agent throughput (claims / day)	5	50	▲ 10×
Touch-time per claim	45 min	3 min	▼ 93%
Processing SLA	72 h	6 h	▼ 91%
Monthly labour spend	₹ 28 L	₹ 4.8 L	▼ 83%
Data errors	6%	< 0.5%	Eliminated