--- title: "Glossary: AI spend, savings and agent action terms · Audon" description: "Definitions of the terms Audon uses for AI spend, savings levers and agent actions: verified saving, prompt caching, idempotency key, prevented loss and more." url: https://audon.ai/glossary last_updated: 2026-11-02 --- # Glossary. The words we use, defined once. Every amount Audon shows carries an evidence class, every saving comes from a named lever, and every agent action has a decision and an outcome. These are the definitions the product, the reports and the API use. ## Spend and savings. The evidence classes every amount carries, from observed spend to a verified saving. - [Observed spend Invoice-aligned, provider-reported cost for the selected UTC days.](https://audon.ai/glossary/observed-spend) - [Anomalous spend Observed spend with features that indicate a possible problem (e.g. a day more than 2.5× its trailing 14-day median).](https://audon.ai/glossary/anomalous-spend) - [Potential saving A modelled reduction if a proposed change works, under stated assumptions.](https://audon.ai/glossary/potential-saving) - [Validated opportunity A scoped, feasible change with supporting evidence and a computable expected benefit, confirmed by a person.](https://audon.ai/glossary/validated-opportunity) - [Verified saving A measured reduction in cost per verified successful task in the latest recorded experiment on production traffic, approved and verified by people other than those who proposed it and recorded the results, with no success-rate drop greater than 2 percentage points, no p95 latency increase greater than 20%, and at least 30 verified successes per window.](https://audon.ai/glossary/verified-saving) - [Cost per verified successful task Cost per verified successful task = Σ attributable run costs ÷ count of verified successful runs.](https://audon.ai/glossary/cost-per-verified-successful-task) - [Cash impact Lower usage reduces future usage-based invoices.](https://audon.ai/glossary/cash-impact) - [Monitoring vs blocking Alerts evaluate complete UTC days of provider billing data, which can lag by hours.](https://audon.ai/glossary/monitoring-vs-blocking) ## Savings levers. The ways AI spend can drop for the same results, as the savings plan names them. - [Service tier Pay the fast or priority surcharge only where someone waits; run work nobody waits on at flex or batch rates.](https://audon.ai/glossary/service-tier) - [Regional processing Drop the regional or US-only processing surcharge where no contract requires it.](https://audon.ai/glossary/regional-processing) - [Model choice Move to a newer, cheaper version of the same model, or route simpler requests to a cheaper model.](https://audon.ai/glossary/model-choice) - [Tests and evals Run tests, evals and CI on a cheaper model, a smaller sample or a nightly schedule.](https://audon.ai/glossary/tests-and-evals) - [Context size Stop resending growing history and stay under long-context price thresholds.](https://audon.ai/glossary/context-size) - [Output and reasoning Cap outputs and set reasoning effort per task, so you pay for the answer, not the padding.](https://audon.ai/glossary/output-and-reasoning) - [Embeddings Embed each document once (skip unchanged content) and run bulk embedding at batch rates.](https://audon.ai/glossary/embeddings) - [Retries and failed runs Stop paying for runs that fail or retry without producing a result.](https://audon.ai/glossary/retries-and-failed-runs) - [Prompt caching Serve repeated prompt prefixes from cache, and stop paying for cache writes that are never read.](https://audon.ai/glossary/prompt-caching) - [Storage and hosted tools Delete stored files, vector stores and caches nobody uses, and cap paid tool calls.](https://audon.ai/glossary/storage-and-hosted-tools) - [Right plan per person Give each person the cheapest plan that covers how they work.](https://audon.ai/glossary/right-plan-per-person) - [Coding model mix Default coding agents to a cheaper model or the tool’s own model pool; keep the premium model for hard tasks.](https://audon.ai/glossary/coding-model-mix) - [Cursor Max mode Keep Max mode for large, multi-file work; normal mode costs less per token for the rest.](https://audon.ai/glossary/cursor-max-mode) - [Overlapping tools People paying for two tools that do the same job.](https://audon.ai/glossary/overlapping-tools) - [Cloud commitments and credits Pay for the same models through AWS, Google Cloud or Azure where a commitment or credits would otherwise go unused.](https://audon.ai/glossary/cloud-commitments-and-credits) - [Reserved capacity Size provisioned throughput to what you use, or move low-use capacity back to pay-as-you-go.](https://audon.ai/glossary/reserved-capacity) ## Agent actions. How an AI agent’s action is decided, executed, verified and counted. - [Protected action A business action an AI agent proposes (a refund, a customer credit, a discount, a subscription change or an order cancel) that runs through Audon.](https://audon.ai/glossary/protected-action) - [Decision What Audon decided about a proposed action.](https://audon.ai/glossary/decision) - [Outcome What actually happened at the provider.](https://audon.ai/glossary/outcome) - [Idempotency key A key of 8–200 characters sent with every execute call, one per intent and reused on every retry of that intent.](https://audon.ai/glossary/idempotency-key) - [Unknown outcome The state of an action when the provider’s answer doesn’t say what happened (a timeout after sending, a 409 or a 5xx).](https://audon.ai/glossary/unknown-outcome) - [Monitoring mode The mode an organization starts in.](https://audon.ai/glossary/monitoring-mode) - [Agent key A key (aa_agent_…) bound to one agent that carries only integrity:execute.](https://audon.ai/glossary/agent-key) - [Observed action An evidence class in Agent Action Control (observed_action).](https://audon.ai/glossary/observed-action) - [Flagged action An evidence class in Agent Action Control (flagged_action).](https://audon.ai/glossary/flagged-action) - [Potentially prevented loss An evidence class in Agent Action Control (potentially_prevented_loss).](https://audon.ai/glossary/potentially-prevented-loss) - [Confirmed prevented loss An evidence class in Agent Action Control (confirmed_prevented_loss).](https://audon.ai/glossary/confirmed-prevented-loss) - [Suspected loss and confirmed loss An evidence class in Agent Action Control (suspected_loss, confirmed_loss).](https://audon.ai/glossary/suspected-loss-and-confirmed-loss) Last updated 2 November 2026 ## Smaller AI bills. Same results, proven. [Start auditFree](https://audon.ai/sign-up) [See a sample report](https://audon.ai/sample-report)