AI Straight-Through Processing for Commercial Claims Automation India

Straight-through processing (STP) in insurance claims means that a claim is assessed, approved, and settled by automated systems without any human reviewing or touching the file. The term originated in banking and capital markets, where transaction processing was automated to remove manual reconciliation steps. In insurance, STP means that from the moment a first notification of loss arrives to the moment a payment is initiated, no human intervenes. The claim travels through intake, validation, coverage determination, quantum assessment, fraud screening, approval, and payment initiation entirely through connected AI and rules-engine systems.

This is a stronger definition than 'claims automation', which often means automating individual steps while humans still review and approve at key junctures. True STP requires that the automation chain has no human checkpoints between FNOL and settlement. In the Indian commercial insurance market in 2026, genuine STP is limited to a subset of claim types and amounts. The majority of automation deployments are semi-automated, where AI handles most steps but a human claims officer performs a final review before approval. The industry distinction between STP and semi-automated processing matters because the economic and regulatory implications are different: STP delivers maximum speed and cost reduction, but also requires higher confidence in each automated decision.

The business case for STP in Indian insurance is grounded in specific numbers. The Insurance Regulatory and Development Authority of India reported in its 2024-25 Annual Report that average claim settlement time for non-life claims was 22 days across the industry, with motor claims averaging 18 days and commercial property claims averaging 31 days. IRDAI's regulations mandate settlement within 30 days of receipt of all required documents, with interest payable at 2% above bank rate for delays beyond this period. STP eliminates the human queue time that accounts for the majority of the gap between document receipt and settlement. Go Digit's internal data, shared at the IRDAI Stakeholder Consultation in January 2026, showed that STP claims settled in an average of 4.2 hours from document receipt, compared to 8.6 days for human-reviewed claims of comparable type and amount.

For Indian commercial insurers, the additional economic driver is the cost of claims handling. A commercial claim handled with full human review involves intake, assignment, document review, coverage determination, reserve setting, negotiation with the surveyor, final approval, and payment processing, with an all-in handling cost estimated at INR 3,500 to 8,000 per claim depending on complexity. An STP claim, once the technology infrastructure is amortised, costs INR 150 to 400 in compute and exception monitoring costs. At scale, the cost differential is material: an insurer processing 50,000 STP-eligible claims per year saves between INR 16 and 37 crore annually in handling costs.

Not all commercial claims are STP candidates. STP viability depends on three factors: the completeness and digital availability of the required evidence; the complexity of the coverage determination; and the amount, where higher amounts warrant human review regardless of other factors. As of mid-2026, STP is viable for a specific subset of commercial claim types in the Indian market.

Motor own-damage claims below defined thresholds are the most STP-mature claim type. For motor fleet policies covering commercial vehicles, own-damage claims where the vehicle attends a network garage, the garage uploads images and a standard damage assessment through the insurer's portal, and the repair estimate falls below the insurer's STP threshold (typically INR 50,000 to 1 lakh depending on the insurer), the claim can be fully processed without human intervention. The damage images are analysed by a computer vision model that estimates repair costs and identifies the damaged components. The policy is checked for coverage, the network garage's assessment is validated against the model's estimate, and if they align within a defined tolerance (typically 15%), payment is authorised to the garage directly.

Simple commercial property claims with complete documentation are STP-viable for relatively small losses. A retail shop or SME unit that suffers a fire or water damage loss, where the surveyor uploads a standard survey report through the insurer's digital portal, the loss amount falls below INR 5 lakh, the policy is straightforward (no endorsements, no co-insurance, no special conditions), and the documentation set is complete, can be processed through STP. The coverage determination is automated through policy parsing; the quantum is taken from the surveyor's report without further negotiation; and payment is initiated on confirmation that the bank account details match the policyholder's KYC record.

Transit and cargo claims with complete shipping documentation are a third STP-viable category. For inland transit claims on road and rail, where the carrier's delivery receipt shows damage, the insured uploads the LR copy, invoice, packing list, and damage certificate, and the claim amount falls below the STP threshold, the claim can be processed without a surveyor visit. The claim documents are validated by NLP models against the policy terms, the loss quantum is computed from invoice value and the damage assessment in the receipt, and payment is initiated after fraud screening.

Claim types that are not STP-viable

Large commercial property claims, engineering claims on operational plants, liability claims, claims involving third-party causation, and any claim where coverage is disputed are not STP candidates. These claims require human judgment on coverage interpretation, surveyor negotiation, legal assessment of liability, or complex quantum computation. Even with advanced AI, the risk of an incorrect automated decision on a INR 5 crore property claim is too high for any Indian insurer to accept without human review. The practical STP ceiling in the Indian commercial market for 2026 is approximately INR 5 to 10 lakh per claim, with most insurers setting lower thresholds while they build confidence in their automated systems.

IRDAI's Regulation 9 of the IRDAI (Protection of Policyholders' Interests) Regulations 2017 requires insurers to settle claims within 30 days of receiving the last required document. Where the insurer needs to conduct an investigation, the period extends to 90 days with a mandatory status communication to the policyholder within 30 days. Non-compliance carries financial consequences: interest at 2% per annum above the Reserve Bank of India's bank rate is payable on delayed settlements, and repeated non-compliance is considered in IRDAI's supervisory rating of the insurer.

IRDAI's 2024-25 enforcement data showed that 14 general insurers received directions for systemic claim settlement delays, with notices citing specific cases where the 30-day window was exceeded by 30 to 90 days without adequate justification. The enforcement context makes STP a compliance tool as well as a cost-reduction tool. An insurer with high STP rates on eligible claims is structurally less likely to breach the settlement timeline because human queue time, the dominant cause of delays, is eliminated for those claims.

The 30-day clock creates an operational pressure that is particularly acute for smaller and mid-sized commercial claims where the document set is complete on arrival. If a commercial motor claim with a complete document set (survey report, repair estimate, invoice, registration copy, driving licence) arrives on Monday morning, and the human claims queue means it is not reviewed until the following week, the insurer has already consumed a significant portion of its 30-day window before substantive processing begins. STP processes the complete document set within hours of arrival, decoupling settlement speed from human capacity.

For claims where documents arrive in stages, STP infrastructure enables document-completeness automation: the system identifies exactly which documents are outstanding, sends automated requests to the insured, and initiates the settlement process immediately when the last document arrives, without waiting for a human to check that the file is complete. The IRDAI clock starts from the date all required documents are received; automating the completeness check and the settlement initiation minimises the time between completeness and settlement. Bajaj Allianz reports that its document-completeness automation on health and small commercial claims has reduced the average time from document completeness to payment initiation from 4.3 days to 6 hours for STP-eligible cases.

The primary risk of STP is not that the AI cannot compute the correct settlement amount. It is that STP creates a faster, wider surface for fraud. A fraudulent claim that would have been caught by a human reviewer who noticed suspicious patterns in the documents, the claimant's history, or the loss circumstances may pass through an automated system that is not configured to check those patterns. Indian general insurers estimate that 3 to 7% of claims by count involve some element of misrepresentation or fraud, with higher rates in motor, health, and SME commercial lines. For STP to be operationally sustainable, every claim must pass through a fraud screening gate before payment authorisation.

The fraud gate for STP-eligible claims in Indian insurance operates across three layers. The first layer is a rules-based screen that checks known fraud indicators without requiring AI inference: policy inception within 30 days of claim (new policy, immediate loss); bank account changed within 7 days of claim submission; claim amount is exactly at or just below the STP threshold (an indicator of threshold gaming); loss date is a Sunday or public holiday in a claim category that shows elevated fraud rates on non-working days; multiple claims from the same location in the same 90-day period. Any positive hit on the rules-based screen redirects the claim to the human fraud investigation queue.

The second layer is an ML-based fraud scoring model trained on historical fraud and legitimate claims. The model receives the structured features from the claim (policy demographics, claim type, amount, timing, surveyor identity, garage identity for motor, claimant's prior claim history from the IIB Industry Loss Database) and produces a fraud probability score. Claims scoring above a defined threshold are redirected to human review. The threshold is calibrated to balance false positive rate (the cost of sending legitimate claims to human review) against false negative rate (the cost of paying fraudulent claims). Indian insurers typically set the threshold to catch approximately 95% of known fraud patterns at the cost of a 5 to 8% false positive rate, meaning 5 to 8% of legitimate STP-eligible claims are redirected to human review as a precaution.

The third layer is document authenticity verification. Survey reports, invoices, and shipping documents submitted through the claims portal are checked for indicators of digital manipulation: metadata inconsistencies, font anomalies, pixel-level editing artefacts in images, and data that does not match the formats expected from the purported source (for example, a truck registration number that does not exist in the VAHAN database, or an invoice date that falls on a day when the vendor was not operational). Computer vision models trained on authentic and manipulated document samples are the primary tool for this layer in Indian deployments.

A claim that clears all three layers, rules screen, ML fraud score, and document verification, proceeds to STP payment authorisation. A claim that fails any layer is routed to a human investigator with the specific flag that triggered the redirect, reducing the human investigator's workload to focused fraud review rather than complete claim assessment. This design means STP and fraud investigation are complementary rather than in tension: STP reduces human load on clean claims, freeing human capacity for the fraud cases that genuinely require investigation.

Go Digit General Insurance and Bajaj Allianz General Insurance represent two different approaches to STP deployment in India, reflecting their different distribution models and technology investment histories.

Go Digit, which operates primarily through digital distribution channels and has invested heavily in API-first claims architecture since its founding in 2016, achieved what it describes as 40% STP rates on motor own-damage claims below INR 75,000 by Q3 2025. The Go Digit STP pipeline integrates its network garage system, which captures repair images, repair estimates, and completion confirmation through a dedicated garage app, with its claims management platform. When a claim arrives through the Go Digit portal, the FNOL parsing agent extracts structured fields, the coverage validation module checks the policy status and peril coverage, the garage assessment is received digitally, and the computer vision model validates the damage claim. If all checks pass and the fraud score is below the threshold, payment to the garage is authorised through the National Payments Corporation of India's IMPS rails without human intervention.

Go Digit's technical infrastructure for STP includes a separate 'STP confidence score' that the system computes for each claim, representing the overall reliability of the automated assessment. Claims with STP confidence above 0.92 proceed automatically; claims between 0.75 and 0.92 proceed with automated settlement but trigger a post-settlement review within 24 hours to catch errors that were not prevented by the pre-settlement checks. Claims below 0.75 are redirected to human review. This confidence-gating approach allows Go Digit to set its STP threshold conservatively while still processing the majority of clean claims without human intervention.

Bajaj Allianz takes a broader but more conservative approach. Its STP deployment spans motor, health, and small commercial property lines, with STP thresholds set lower than Go Digit's (INR 50,000 for motor, INR 3 lakh for small commercial property). Bajaj Allianz reports STP rates of 25 to 30% across its eligible claim volume as of Q1 2026, with the lower rate reflecting its broader product mix and more conservative fraud screening thresholds rather than less sophisticated technology. Bajaj Allianz's STP infrastructure includes a real-time connection to the Insurance Information Bureau of India (IIB) database, allowing the fraud scoring model to access the claimant's full loss history across insurers in real time during the STP assessment.

Bajaj Allianz's post-STP audit process is a model for the industry: every STP claim is included in a stratified monthly sample where human claim officers review the automated decision in retrospect. The audit sample is biased toward claims near the fraud score threshold and toward claim types with historically higher fraud rates. Audit findings feed back into the fraud model retraining cycle on a quarterly basis, and any systematic errors in the automated assessment result in a temporary STP pause for that claim sub-type while the model is recalibrated.

The Indian non-life insurance industry's STP rate across all claim types was approximately 12 to 15% by volume in FY2025-26, according to industry data presented at the GI Council annual conference in February 2026. This aggregate figure masks significant variation: motor own-damage STP rates at digital-first insurers reached 35 to 45%, while commercial property and liability lines remained at 3 to 8%.

The trajectory over the next three years points toward STP rates of 30 to 40% across all claim types as digital document submission becomes more prevalent, surveyor platforms become more data-rich, and fraud models mature on larger training datasets. The key bottleneck is not AI capability but infrastructure readiness at the point of claim origination. STP requires that all claim inputs arrive in structured, machine-readable form. A survey report submitted as a scanned PDF requires OCR and NLP extraction before it can be processed, and extraction errors create STP failures. A survey report submitted through a digital surveyor platform as a structured JSON object with image attachments can be ingested by the STP system with near-100% reliability.

The IRDAI Bima Sugam platform, operational in full form from 2026, is expected to accelerate STP adoption by creating a common digital infrastructure for claim submission, policy lookup, and identity verification. When the policy exists on Bima Sugam and the claim is submitted through the Bima Sugam claims module, the STP system can retrieve verified policy data from the platform rather than relying on self-reported policy details from the claimant. The identity verification layer on Bima Sugam (linked to Aadhaar and PAN) also reduces the risk of claim submission by ineligible parties, removing one category of fraud that currently requires human attention.

Survey digitalisation is the second infrastructure investment that will determine STP uptake in commercial lines. The Insurance Brokers Association of India and the Surveyor and Loss Assessors Association have been working with IRDAI since 2024 on a digital survey report standard that would allow structured data from surveyor assessments to flow directly into claims systems via API. If adopted widely, this standard would make commercial property and engineering claims with amounts up to INR 25 lakh STP-viable, dramatically expanding the addressable STP market. A pilot of the digital survey report standard is underway with 12 insurers and 200 panel surveyors as of April 2026.

Limits of STP in commercial lines

The structural limits of STP in Indian commercial insurance will persist even as technology and infrastructure mature. Claims involving third-party liability determination, disputes over policy interpretation, losses where the cause is contested (fire versus electrical failure, storm versus manufacturing defect), claims requiring expert inspection of specialised equipment, and all claims above defined financial thresholds will continue to require human judgment. The realistic long-term STP ceiling for Indian commercial insurance is approximately 40 to 50% by claim count but only 15 to 25% by claim value, because the largest and most complex claims will always fall outside STP eligibility. The commercial insurance industry's investment in STP should be calibrated against this ceiling: automating the high-volume, low-complexity end of the claim portfolio frees underwriting and claims talent for the complex work that genuinely requires human expertise.