AI Partners Usage Management

Overview#

An agency runs dozens of AI-powered workflows each day: document summarisation, entity extraction, hypothesis testing, and chat-based analysis. By the middle of the month, they've burned through 80% of their AI budget with two weeks still to go. Without usage management, the options are either cut off AI access or overspend. Neither is acceptable in an operational environment.

The AI Partners Usage Management module prevents that situation. It tracks every AI API call in real time, calculates costs accurately against per-provider pricing, enforces budget limits, and gives decision-makers the data they need to optimise how AI is used across the platform.

Key Features#

Real-Time Token Tracking: Automatic recording of every AI API call with input and output token counts for complete usage visibility.
Accurate Cost Calculation: Database-driven pricing model with per-token rates ensures precise cost attribution across all AI providers.
Budget Management: Set monthly budgets per organisation with automatic alerts at configurable thresholds (80%, 90%, and exceeded) and optional hard spending limits.
Rate Limiting: Hourly and daily request quotas per AI provider prevent runaway usage and protect shared resources.
Analytics Dashboard: Comprehensive usage insights broken down by AI provider, operation type, investigation, and time period for informed decision-making.
Fallback Strategy: Priority-based multi-provider support with automatic failover ensures continuity when a provider is unavailable or rate-limited.
Data Export: Generate usage reports in CSV, JSON, and PDF formats for compliance, auditing, and financial reporting.
Per-Provider Configuration: Enable, disable, and configure each AI provider independently per organisation with custom priorities and rate limits.
Cost Projections: Estimate month-end costs based on current usage trends to help organisations plan and manage their AI budgets proactively.
Programmable API Access: Full API support for querying usage data, managing budgets, updating configurations, and exporting reports.

Use Cases#

Government agencies with fixed annual AI budgets set hard monthly spending limits per department, ensuring that operational AI usage never exceeds approved appropriations regardless of workload spikes.

Law enforcement units attribute AI costs to individual investigations, giving supervisors visibility into which cases are consuming the most analytical resources and allowing budget reallocation when needed.

Financial crime teams run compliance audits by exporting detailed usage records showing exactly which AI operations were executed, when, and at what cost, satisfying internal governance review requirements.

Intelligence organisations managing multi-provider AI deployments use the provider performance dashboard to make informed decisions about which providers deliver the best cost-to-quality ratio for specific analysis types.

Cost Control: Set monthly budgets and receive automatic alerts as spending approaches thresholds, with optional hard limits to prevent overspend.
Usage Optimisation: Analyse which AI operations consume the most tokens and cost the most, then optimise workflows to reduce unnecessary AI calls.
Compliance Reporting: Export detailed usage records for financial audits, regulatory compliance, and internal governance reviews.
Capacity Planning: Use trend analysis and cost projections to forecast future AI spending and plan budgets for upcoming periods.

Integration#

The AI Partners Usage Management module connects with other Argus modules:

AI Partner Platform: Tracks token consumption for all AI partner conversations and knowledge base operations.
Document Analysis: Records usage and costs for AI-powered document processing and analysis operations.
Smart Fields: Monitors token usage for AI-assisted field enrichment and data extraction.
Investigation Management: Attributes AI costs to specific investigations for per-case cost tracking.
Administration: Budget management integrates with organisational administration for billing and governance.

Open Standards#

GraphQL (June 2018 specification): the entire module API surface is exposed as a typed GraphQL schema, with queries for usage analytics and cost summaries, and mutations for budget updates, partner settings, and usage export.
JSON (RFC 8259): all API responses, internal usage event metadata, and one of the three export formats are serialised as JSON, with ISO 8601 datetime strings used throughout.
JSON Web Token (RFC 7519) / JSON Web Key Set (RFC 7517): access control relies on RS256-signed JWTs validated against a JWKS endpoint; every query and mutation requires a verified bearer token.
OAuth 2.0 (RFC 6749): the IsAuthenticated permission class enforces OAuth 2.0 bearer-token authorisation on all GraphQL operations, gating usage tracking and budget management behind a valid access token.
CSV (RFC 4180): usage records can be exported in RFC 4180-compliant comma-separated-values format, with a header row and sanitised field values, for use in financial and compliance tooling.
ISO 8601: all timestamps recorded in usage events and returned in API responses are encoded in ISO 8601 format, ensuring unambiguous date and time interchange across systems and export consumers.
PDF (ISO 32000): the export feature supports PDF as a named output format, allowing organisations to generate human-readable usage and cost reports suitable for governance and audit submissions.

Last Reviewed: 2026-02-05 Last Updated: 2026-04-14