Connecting insurers, banks, regulators, and ecosystem partners through a shared network for policy verification, fraud detection, and seamless claims.
Designed for the institutions that govern insurance trust
Not long ago, moving money between banks in Indonesia was slow, bound to office hours, and stitched together through fragmented bilateral links. In December 2021, Bank Indonesia launched BI-Fast: one shared national rail for real-time, round-the-clock interbank transfers. Today more than 120 banks plug into that same infrastructure, and a transfer that once took a day now clears in seconds.
Insurance today looks like banking did before BI-Fast. Every insurer sits in its own silo, the same verification is repeated by each party, and fraud that crosses company lines stays invisible. There is no shared rail.
Bank Indonesia did not replace the banks; it gave them a common rail so trust could move between them. IDRN does the same for insurance, without ever pooling raw customer data in one place.
// BI-Fast, Bank Indonesia national fast-payment rail, launched Dec 2021; 120+ participating banks. IDRN applies the same shared-infrastructure model across the policy-to-claim lifecycle.
Every insurer assesses policies and claims inside its own data silo. A policy's authenticity is hard to prove to outsiders, the same policyholder can file the same loss with two companies undetected, and supervisors see the industry only through delayed periodic reports. Five structural failures compound one another.
A policyholder's policy and claim history at one insurer is invisible to every other. Double claims on a single loss are practically undetectable.
Policy files, damage photos and workshop estimates live on each party's own systems, and can be altered without leaving any auditable trace.
Suspicion rests on an adjuster's intuition. There is no explainable risk standard that can be compared across cases and defended.
Regulators receive a delayed snapshot. Cross-company fraud patterns and RBC pressure only surface after losses accumulate.
Because the system cannot quickly separate honest claims from suspicious ones, all claims are processed slowly, and premiums carry the cost of fraud.
The best fraud engine is useless if its data can be manipulated. The most secure ledger is useless if nothing flags which claim to inspect. The problem demands both, together.
IDRN now spans the entire insurance lifecycle, from the moment a policy is issued to the moment a claim is resolved, through two products that share the same integrity backbone and oversight.
Every issued policy receives an IDRN Number, a verification QR, and a SHA-256 digital fingerprint anchored to the ledger. A policy authenticity can be proven by anyone, and premium growth is visible in real time.
Every claim is scored by the IDRN-RS engine (0-100) with explainable factors, its fraud patterns mapped, and its pressure on RBC monitored. The decision always stays with the insurer.
Every IDRN policy carries a unique QR. A policyholder, an agent, or anyone can scan it to verify the policy's authenticity against the network, with no login. It either matches or it does not, in a single scan.
// The SHA-256 fingerprint matches the Merkle root anchored on the ledger. Scan this QR to open the verification portal.
IDRN is not an insurer and does not take over decisions. It provides three things: an explainable risk score for every claim, integrity proof for every document, and an audit trail for every decision, powered by two complementary engines.
IDRN-RS is a tool, not a judge. It reads and summarises every incoming claim in under a second, maps the patterns that signal fraud, and returns a 0-100 score that is the sum of weighted, human-readable factors. Never a black box: every point traces back to the factor that produced it, and the call stays with the adjuster.
A permissioned blockchain, not a currency or token, but an integrity machine. Every document is fingerprinted with SHA-256, Merkle-batched, and immutably anchored on-chain to an append-only distributed ledger, secured by cryptographic consensus across three independent validator nodes.
The AI is a tool that reads every claim, summarises it, and maps the patterns of fraud. The blockchain is the backbone that keeps what it finds impossible to forge. The decision always stays with the insurer.
IDRN never decides a claim. It speeds the workflow so claim decisions are faster and more transparent, trust rises, and insurance protection reaches more Indonesians. This is shared national infrastructure, built together by many parties for the good of the public, owned by no single one.
A claim scoring 72/100 isn't a verdict. It's an argument. Adjusters, supervisors, and even the customer can see exactly where every point came from, then act within a defined risk band.
Total 72/100. Each score carries a model confidence reflecting how complete the supporting data is. Illustrative decomposition per the IDRN whitepaper.
The fast-track threshold is a business policy with measurable consequences. A higher bar raises automation and cuts cost, but it also raises mis-payment risk. IDRN simulates every threshold against the real score distribution before it ships. Production changes require two-officer approval, logged to the audit trail.
From the moment a policy is issued to the moment a claim is resolved, every step is scored, anchored, and logged.
A policy is issued; its fingerprint is computed and it receives an IDRN Number and verification QR.
Anyone scans the QR or enters the IDRN Number to confirm the policy is genuine. Status: VERIFIED.
Fingerprints enter the anchoring queue, validated by IDRN, OJK and AAUI.
IDRN-RS computes the score in under a second and checks for cross-insurer duplicates.
The adjuster reviews the decomposition then approves, rejects or escalates, with a mandatory written reason logged permanently.
If a cross-insurer double claim is detected, supervisor and association are auto-notified.
The policyholder follows every stage in real time; a rejection comes with a readable basis and a 30-day right to appeal.
IDRN connects the whole ecosystem in one shared trust layer, giving each role exactly the access it needs, with every session recorded on the audit trail.
Issue policies, register claims, receive risk scores, process payment, and monitor portfolio and high-risk entities.
Review claims via score decomposition and decide with a mandatory, audited written reason.
Aggregate cross-insurer monitoring, real-time RBC, double-claim alerts, full audit-trail access, and ledger validator.
Industry policy and claim registry, watchlist and advisory distribution, data-schema standards, and ledger validator.
Verify policy authenticity via QR, coverage transparency, real-time claim tracking, and the right to appeal any rejection.
Each insurer registers a claim signature (a combined hash of the policyholder's identity, the insured object, and the loss-date range) into a registry co-governed with the association.
When two signatures from different insurers collide, the network flags the double-claim group and raises factor F3 on both claims, without either party ever seeing the other's file. Detailed confirmation happens only after a signal, through official recorded channels, with the supervisor as witness.
This is why the problem cannot be solved by one company alone, and why the solution must never be a merged customer database.
Signature collision detected. Double-claim group flagged on both claims. Raw files never exchanged.
// SHA-256 is one-way; identity cannot be reconstructed from the hash. Compliant with UU No. 27/2022 (PDP) by architecture, not just procedure.
Claim decisions remain entirely with licensed insurers under OJK rules. IDRN is a technology infrastructure provider, and privacy is enforced architecturally.
Personal data and raw documents never leave the insurer's systems. Only one-way fingerprints, risk signals and dedup metadata cross the network.
No single party, not even the IDRN operator, can alter the network's records. Every batch anchor is signed by three independent validators.
AI recommends; it never decides. Every approval, rejection or escalation is made by a human with a mandatory written reason, recorded permanently.
Validator signing keys are protected by hardware security modules with periodic rotation, so integrity proofs arrive in real time, not by report.
The natural adoption route: an OJK regulatory sandbox that begins with general insurance and the AAUI, shared data-schema standardisation, then gradual expansion to every insurance line.
Adding validators, changing factor weights, or setting thresholds all follow a multi-party approval mechanism, fully recorded on the audit trail.
The AI answers which claim to look at. The blockchain answers a different question: can this data and decision be trusted, unaltered. That second question is why IDRN needs a ledger, not just a database.
This is the real reason for the ledger, not hype and not a token. A permissioned blockchain is honest because of how it is built: records can only be added and never rewritten, they are replicated to independent parties, and they are valid only when those parties validate them together. No single party, not even IDRN, can change the past on its own.
The first insurer to join gets a scoring engine. The tenth gets a scoring engine that sees ten times the signals.
All figures shown are design targets for a demonstration environment, drawn from the IDRN whitepaper, not audited production results. Actual outcomes depend on each insurer's portfolio and must be verified by a third party. IDRN describes a functional prototype architecture; a full machine-learning scoring model requires cross-insurer production data available only once the network is running.
Today IDRN-RS is an explainable tool. The ambition is far bigger: the connective infrastructure that an entire industry, and eventually other markets, are built on.
The near-term goal is to become the national standard: the default trust layer Indonesian insurance is built on, from policy to claim, endorsed within the OJK framework.
As every participant connects, the ecosystem tightens. Insurers, banks, reinsurers, regulators, and policyholders on one network, so a fraud signal seen by one becomes protection for all, and fraud is pushed down across the whole market, not just inside a single company.
The same shared-infrastructure model that BI-Fast proved for payments can reach beyond Indonesia, giving other markets a verifiable trust layer of their own.
Weighted factors F1–F5 with a confidence level, so every claim can be summarised and defended, factor by factor.
Machine-learning models trained on cross-insurer production data, with sharper value-anomaly detection and fuzzy matching that resists signature evasion.
Adopted as the default trust layer across Indonesian insurance: every line, every participant connected, and fraud pushed down across the whole ecosystem.
The same shared-infrastructure model extended to regional and global insurance markets, one verifiable rail at a time.
Banking got BI-Fast. Insurance is still waiting for its trust layer, and it only gets built once. IDRN is that infrastructure, from policy to claim, with regulation on its side and network effects that grow stronger with every member. Come in early: help set the national standard, drive fraud down across the whole market, and share in the category it creates. A window like this does not open twice.