Claims intake went digital, and the routine ones started settling themselves

The client is a regional insurance group writing motor, property and medical lines across MENA markets, with a branch network, a broker channel and a claims operation that had grown up around paper. A claim began as a phone call or a branch visit. Photos of a damaged bumper arrived by email or messaging apps. Intake clerks printed attachments, assembled physical files and re-keyed details into the core policy administration system that underwriting had already captured at policy issue.

The model held while volumes were modest and adjusters knew their claimants. It stopped holding when motor volumes grew and digital-first customers began comparing the group's claims experience to their banking apps. Every claim, whether a windscreen chip or a warehouse fire, followed the same manual path, so trivial claims consumed the same adjuster attention as complex ones. Files sat in inboxes, fraud checks depended on what a handler happened to remember, and approved settlements waited for finance to assemble bank payment runs by hand. Regulators across the group's markets were tightening expectations on settlement timeliness and complaint handling, and the operation had no reliable way to show where any given claim stood.

Kentro was engaged to build a claims automation layer over the existing core systems, not to replace them. The scope covered digital claims intake for photos and documents, rules-based triage, straight-through processing for simple claims, fraud flagging, an adjuster workbench, and integration into the group's policy administration systems and payment rails. Kentro brought the delivery patterns from Claimoto, its own insurtech platform, and adapted them to the group's lines of business and market-by-market regulatory requirements.

Claims ran on effort and memory. The group's people knew how to settle claims; the process gave them no leverage.

• Every claim took the same path.: A windscreen chip and a warehouse fire entered the same manual queue. Nothing distinguished routine from complex until an adjuster opened the file, so scarce judgment was spent on claims that needed none.
• Intake was paper and inboxes.: Photos and documents arrived over email and messaging apps, were printed and rescanned, and intake staff re-keyed everything into the core policy system. The claim file was scattered across systems and drawers.
• Fraud screening depended on memory.: Duplicate submissions and recycled damage photos were caught only when an experienced handler happened to recognize them. There was no systematic check applied to every claim.
• Settlement ended in a manual payment run.: Approved claims waited for finance to prepare bank files by hand. Even a fully approved simple claim sat unpaid until the next run was assembled.

Kentro built the platform as an event-driven service layer between customer channels and the group's core policy administration systems. A claim moves through the system as a stream of events (submitted, documented, triaged, adjudicated, settled), with each service subscribing to the events it needs and an anti-corruption layer isolating the platform from legacy core interfaces.

• Digital intake built for photos and documents.: Mobile-responsive notice-of-loss flows with guided photo capture, direct uploads to object storage, and EXIF metadata retention for later verification. OCR-based classification indexes registration cards, police reports and repair invoices against the claim automatically.
• Triage as editable decision tables.: Triage logic lives in a decision engine as DMN-style decision tables: coverage verification against the policy system, reserve suggestion, and routing into straight-through, fast-track or adjuster queues. Claims leadership adjusts rules and thresholds without a software release.
• Straight-through processing for simple claims.: In-scope motor and property claims get automated coverage and deductible checks, auto-adjudication, and a generated payment instruction, with every automated decision logged for audit. A clean claim moves from submission to settlement without a human touch.
• Fraud flags at intake and adjudication.: Perceptual image hashing catches reused damage photos, metadata checks compare device, location and capture time against the reported loss, and watchlist and prior-claims matching runs on every submission. Flagged claims route to investigators instead of the straight-through path.
• Adjuster workbench on an integration spine.: A queue-based workbench presents the complete indexed claim file with SLA timers and a full audit trail. REST APIs wrap the legacy core interfaces, and payment instructions flow to the group's banks over their corporate payment APIs with IBAN validation before release.

The engagement changed what the claims organization spends its time on.

• Routine claims settle without manual handling.: Straight-through claims move from submission to payment instruction without an adjuster opening the file. Intake staff no longer re-key claim data into the core system.
• Adjusters work judgment, not paperwork.: The workbench presents a complete, indexed claim file, so adjuster time goes to liability, quantum and negotiation instead of chasing documents across inboxes.
• Fraud review became systematic.: Every claim passes the same duplicate, metadata and watchlist checks. Investigators start from flagged evidence rather than a handler's hunch.
• The business changes its own rules.: Triage and routing live in decision tables that claims leadership can adjust directly, so product and process changes no longer wait on an IT release. The audit trail gives regulators a clear record of how every claim was decided.