specification: FinOps Framework
specificationVersion: '1.0'
schema: https://www.finops.org/framework/
provider: Waxell
providerId: waxell
created: '2026-05-06'
modified: '2026-05-06'
reconciled: false
notes: >-
  Waxell is currently in beta with all commercial pricing routed through a
  sales call; the public site does not publish per-meter prices. Meters and
  charge categories below model the platform's documented telemetry surface
  (Observe runs, LLM calls, governance policy evaluations, MCP tool calls,
  workflow runtime executions) but unit prices are not yet known. Set
  reconciled to true after pricing is published or confirmed in a customer
  agreement.
tags:
  - FinOps
  - FOCUS
  - AI Agent Governance
  - Observability
description: >-
  FOCUS-aligned FinOps shape for Waxell: a usage-based AI agent governance and
  observability platform metering ingested LLM calls, governance policy
  evaluations, MCP tool calls, and durable runtime executions, with subscription
  add-ons for enterprise governance and seats.
sources:
  - https://waxell.ai/
  - https://waxell.ai/get-access
  - https://waxell.ai/docs/observe/features/cost-management
  - https://waxell.ai/docs/observe/api/endpoints
  - https://waxell.ai/docs/security
alignedWith:
  framework: FinOps Foundation Framework
  frameworkUrl: https://www.finops.org/framework/
  dataSpec: FOCUS
  dataSpecVersion: '1.3'
  dataSpecUrl: https://focus.finops.org/focus-specification/v1-3/
publisherName: Callsine, Inc. (dba Waxell)
serviceCategory: AI Agent Governance and Observability
billingModel:
  pricingCategory: Pay-As-You-Go + Subscription
  billingFrequency: Monthly
  billingCurrency: USD
  chargeCategories:
    - Usage
    - Purchase
    - Tax
    - Adjustment
    - Credit
focusColumns:
  ServiceName: Waxell
  ServiceCategory: AI Agent Governance and Observability
  ProviderName: Waxell
  PublisherName: Callsine, Inc.
  InvoiceIssuerName: Callsine, Inc.
  BillingCurrency: USD
  ChargeCategory: Usage
  RegionId: us-east-1, eu-west-1
meters:
  - name: agent_runs
    unit: request
    aggregation: count
    dimensions:
      - agent_name
      - tenant
  - name: llm_calls_ingested
    unit: request
    aggregation: count
    dimensions:
      - provider
      - model
      - agent_name
  - name: tokens_input
    unit: token
    aggregation: sum
    dimensions:
      - provider
      - model
  - name: tokens_output
    unit: token
    aggregation: sum
    dimensions:
      - provider
      - model
  - name: policy_evaluations
    unit: request
    aggregation: count
    dimensions:
      - category
      - decision
  - name: mcp_tool_calls
    unit: request
    aggregation: count
    dimensions:
      - mcp_server
      - tool_name
  - name: durable_runtime_executions
    unit: instance-hour
    aggregation: sum
    dimensions:
      - workflow_name
  - name: data_retention
    unit: GB-month
    aggregation: avg
    dimensions:
      - data_class
  - name: seats
    unit: seat
    aggregation: max
    dimensions:
      - role
principles:
  - name: Visibility
    description: Use the Waxell control plane dashboard at https://waxell.dev plus the Observe REST API (`/runs`, `/llm-calls`, `/spans`, `/events`) and the `estimate_cost()` SDK helper to see token, dollar, and policy-decision consumption per agent and per user.
  - name: Allocation
    description: Tag every run with `agent_name`, `session_id`, and `user_id` plus free-form metadata via WaxellContext to attribute LLM and governance spend to the team or product owning the agent.
  - name: Optimization
    description: Use `model-costs` overrides to reflect negotiated rates from upstream LLM vendors; use Cost Management governance policies to set budgets that warn, throttle, or block when an agent or tenant approaches a spend ceiling; use the prompt manager to roll back regressions that inflate token usage.
  - name: Accountability
    description: SOC 2 Type II is in progress; immutable audit trails (`/events/`) plus the kill-switch policy and human-in-the-loop approval handlers (Slack, webhooks) give a single owner the levers to halt or escalate spend in real time.