REST API
Your core system, DMS or workflow triggers analyses and receives structured results – no extra screen for claims handlers.
Insurance AI Workbench
Ready-to-use AI for claims, underwriting, contracts and actuarial.
The central AI platform for property, life and health insurers: field-tested, integrated into your core systems through APIs and expandable with specialist skills – with versioning, roles and a complete audit history.
A controlled AI analysis engine – not a chatbot.
Powered by 9elf26.ai
Define the insurance workflowControlled process
Manage prompt and model versionsVersioned
Connect approved knowledge sourcesAccess controlled
Run AI-supported analysisRepeatable
Apply the review ruleStraight-through or expert approval
Return approved output through APIAudit history
Prompt management · Multi-model connection · Knowledge management · Role-based access · Quality review · Full audit trail · API integration
Up to40%lower costs
Up to70%less processing effort
Achievable with the Insurance AI Workbench. Results vary by process and implementation.
How the Workbench is used
Where your teams already work.
Most work with the Workbench doesn’t happen in the Workbench: your systems, AI agents and interfaces call it. The console is where experts configure what runs behind the scenes.
MCP
Through the Model Context Protocol, AI assistants and agents use Workbench skills as tools – as approved, versioned skills.
Your interfaces
Results appear in the applications your teams already use: claims software, portals or intranet.
Workbench console
Where insurance experts configure workflows, manage prompt and model versions and monitor results.
Ready for every line of business
Four areas. One central platform.
For property, life and health insurers – field-tested and tailored to your processes.
Underwriting & risk assessment
Risks are assessed automatically using defined decision logic and integrated into your pricing models – with results in real time.
Claims management
Claim notifications are analysed automatically, relevant data is extracted and next steps are suggested – faster, cheaper, more consistent.
Customer service & contracts
Customer and contract data is processed via APIs with your CRM and core system; requests and changes are written directly into them.
Actuarial
Portfolio and claims data is evaluated automatically – for forecasts and reports without manual data preparation.
Operating modes
Fully automated, interactive or batch.
Set per process – with review rules that decide when an expert takes over.
Straight-through processing
Standard cases run fully automated. Exceptions are routed to your experts.
Interactive
Claims handlers work online with the AI, review suggestions and approve results.
Batch
Large portfolios are processed in bulk – for analyses, forecasts and reports.
Why a workbench
Repeatable, not random.
A chatbot answers questions. A workbench runs approved analyses – the same way every time, traceable and under your experts’ control.
Approved workflows, not free-form prompts
Insurance experts define and test what the AI does. Only approved versions run in production.
Versioned and traceable
Prompts, models and knowledge sources are versioned. Every run leaves an audit trail.
Results where the work happens
Analyses are triggered through APIs and return to your core system, DMS or workflow.
Capabilities
What the Workbench brings.
- Prompt management and versioning
- Multi-model connection, cloud or on-premise
- Knowledge management for policy terms and guidelines
- Role-based access
- Quality review by insurance experts
- Full audit trail
- API integration with core systems
- Less dependence on a single model vendor
- Expandable with specialist skills – each skill adds technical depth and auditability
- Built for AI Act requirements: human oversight, logging, documentation
One workflow, both modules
From incomplete notice to an expert decision.
FNOL completeness shows how Capture and the Workbench operate as one controlled process.
Claim enters
A notice arrives through the existing process.
Completeness check
The Workbench checks whether all FNOL information is present.
Capture triggered
Missing evidence starts a guided capture.
Evidence returned
The customer submits photos, video or documents.
Analysis prepared
The evidence is structured for the workflow.
Expert review
An expert reviews exceptions and decides.
Result returned
Result and audit history return to the core system.
Use cases
33 starting points for your insurance processes.
From AI damage analysis to contract changes and actuarial analysis – every use case can be the starting point for a pilot.
AI damage analysis
Automated analysis of photos and statements at FNOL and inspection – returns damage type, extent and a first assessment.
Discuss this use caseCoverage check
Checks automatically whether the reported loss is covered by the policy and flags borderline cases for manual review.
Discuss this use caseIntelligent email triage
Classifies incoming email from shared mailboxes – from customers, intermediaries, repair shops, police or fire services – and routes it to the right department and line of business. No message is left sitting in the wrong mailbox.
- Extracts policy and claim numbers, VIN and licence plate
- Short summary, classification and next step for every email
- Falls back to customer service when no specialist team fits
Fraud assessment
Checks statements and media for anomalies and points the claims handler to them.
- Plausibility check of statements
- Media forensics for photos and video
- Alert to the claims handler when something looks off
Requests for missing documents
Detects missing information, documents or photos in claims and applications, drafts the right request to the customer, intermediary or loss adjuster and tracks whether it arrives.
Discuss this use caseClaim summary
Combines photos, videos, documents and spoken statements into a structured claim overview.
Discuss this use caseBook a demo
Bring one process. We’ll show you where Aestimo fits.
A specific use case is the best starting point.