Advanced Analytics Platform

Overview#

Intelligence organisations processing hundreds of reports a day cannot afford to run each one through a manual analysis workflow. The Advanced Analytics Platform provides the orchestration layer that routes those reports through the right AI models, enforces cost controls, tracks every token spent, and delivers results back into the investigation workflow automatically. When a financial crime unit needs multi-model consensus on a high-stakes suspicious activity report, the platform runs simultaneous analyses across multiple AI language models, then aggregates the findings with confidence scoring before surfacing them to the analyst.

The platform supports multi-LLM consensus validation, browser-based Python analytics, and reusable template workflows accessible through standardised APIs, all with organisation-level isolation in a multi-tenant environment.

Key Features#

• Multi-LLM provider orchestration across multiple AI language models with automatic failover and cost-aware routing
• Dynamic provider selection enabling multi-model consensus validation for high-stakes investigations
• Client-side Python analytics execution via browser-based runtime with full scientific computing libraries at zero server compute cost
• Reusable analytics template management encapsulating organisation-specific methodologies with version control and security review
• Real-time job orchestration with status tracking, progress updates, and multi-provider result aggregation
• Granular cost tracking and budget management with per-analysis cost attribution, token usage monitoring, and organisational spending limits
• Real-time subscriptions for server-push job progress updates without polling overhead, delivered via GraphQL subscriptions
• Batch operations for simultaneous multi-analysis requests
• Template marketplace for cross-organisational sharing of validated analytical workflows

Use Cases#

• Financial Crime Network Mapping: Financial intelligence units analyse suspicious activity reports using multi-LLM consensus, combining entity extraction, graph construction, and centrality analysis to identify money laundering networks across jurisdictions.
• Law Enforcement Entity Extraction: Fusion centres process hundreds of daily intelligence reports using browser-based entity extraction templates, operating effectively even during network degradation.
• Regulatory Compliance Template Marketplace: Multi-national organisations standardise analytical workflows across subsidiaries with centralised template governance, budget controls, and cost attribution for audit purposes.

Integration#

• Connects with investigation platforms and case management systems through typed APIs
• Real-time subscriptions for live job progress monitoring via GraphQL
• Integrates with SIEM platforms for analysis completion event triggering
• Compatible with business intelligence tools for executive dashboard generation
• Webhook endpoints for posting results to collaboration platforms and ticketing systems
• Multi-tenant data isolation with role-based access control and comprehensive audit logging
• PostgreSQL primary data store for all job state, cost records, and result persistence

Open Standards#

• GraphQL (June 2018 Specification): the entire analytics API surface, including multi-LLM job orchestration, template management, session tracking, and real-time job-progress updates, is exposed exclusively through GraphQL queries, mutations, and subscriptions implemented with the Strawberry framework.
• WebAssembly (W3C WebAssembly Core Specification): client-side Python analytics execution uses Pyodide, which compiles CPython to WebAssembly and runs it in the browser, eliminating server compute cost for analyst-driven data transformations.
• JSON Schema (IETF draft-07 / 2020-12): every analytics template carries an expected_input_schema and an expected_output_schema expressed as JSON Schema documents, enabling automated validation of inputs and outputs for reusable analytical workflows.
• JSON Web Token / RFC 7519: all GraphQL and REST analytics endpoints are protected by RS256-signed JWTs verified against a JWKS endpoint; the IsAuthenticated permission class enforces this on every query and mutation.
• ISO 8601: all temporal data across job state, event sourcing records, and KPI route responses is serialised as ISO 8601 timestamps, with explicit UTC normalisation throughout.
• WebSocket / RFC 6455: the event sourcing service broadcasts domain events (case, alert, evidence, entity) to browser clients in real time over WebSocket connections hosted on Cloudflare Durable Objects, underpinning live dashboard updates.
• AMPDS (Advanced Medical Priority Dispatch System): the NAS KPI analytics routes apply AMPDS priority bands (Echo, Delta, Charlie, Bravo, Alpha) to incident data to calculate SLA compliance and projected-breach analytics for emergency medical dispatch performance reporting.

Last Reviewed: 2026-02-23 Last Updated: 2026-04-14