Glossary. The words we use, defined once.
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.
Anomalous spend
Observed spend with features that indicate a possible problem (e.g. a day more than 2.5× its trailing 14-day median).
Potential saving
A modelled reduction if a proposed change works, under stated assumptions.
Validated opportunity
A scoped, feasible change with supporting evidence and a computable expected benefit, confirmed by a person.
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.
Cost per verified successful task
Cost per verified successful task = Σ attributable run costs ÷ count of verified successful runs.
Cash impact
Lower usage reduces future usage-based invoices.
Monitoring vs blocking
Alerts evaluate complete UTC days of provider billing data, which can lag by hours.
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.
Regional processing
Drop the regional or US-only processing surcharge where no contract requires it.
Model choice
Move to a newer, cheaper version of the same model, or route simpler requests to a cheaper model.
Tests and evals
Run tests, evals and CI on a cheaper model, a smaller sample or a nightly schedule.
Context size
Stop resending growing history and stay under long-context price thresholds.
Output and reasoning
Cap outputs and set reasoning effort per task, so you pay for the answer, not the padding.
Embeddings
Embed each document once (skip unchanged content) and run bulk embedding at batch rates.
Retries and failed runs
Stop paying for runs that fail or retry without producing a result.
Prompt caching
Serve repeated prompt prefixes from cache, and stop paying for cache writes that are never read.
Storage and hosted tools
Delete stored files, vector stores and caches nobody uses, and cap paid tool calls.
Right plan per person
Give each person the cheapest plan that covers how they work.
Coding model mix
Default coding agents to a cheaper model or the tool’s own model pool; keep the premium model for hard tasks.
Cursor Max mode
Keep Max mode for large, multi-file work; normal mode costs less per token for the rest.
Overlapping tools
People paying for two tools that do the same job.
Cloud commitments and credits
Pay for the same models through AWS, Google Cloud or Azure where a commitment or credits would otherwise go unused.
Reserved capacity
Size provisioned throughput to what you use, or move low-use capacity back to pay-as-you-go.
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.
Decision
What Audon decided about a proposed action.
Outcome
What actually happened at the provider.
Idempotency key
A key of 8–200 characters sent with every execute call, one per intent and reused on every retry of that intent.
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).
Monitoring mode
The mode an organization starts in.
Agent key
A key (aa_agent_…) bound to one agent that carries only integrity:execute.
Observed action
An evidence class in Agent Action Control (observed_action).
Flagged action
An evidence class in Agent Action Control (flagged_action).
Potentially prevented loss
An evidence class in Agent Action Control (potentially_prevented_loss).
Confirmed prevented loss
An evidence class in Agent Action Control (confirmed_prevented_loss).
Suspected loss and confirmed loss
An evidence class in Agent Action Control (suspected_loss, confirmed_loss).
Last updated