Skip to content
All workshops

Private, hands-on workshops with the toolkit maintainers

FinOps for AI

A map of your AI costs, and what your FOCUS data can and cannot answer about them.

You bring

Your AI workloads, assistant licensing, usage data, and the people accountable for the bill.

We build together

A shared map of where your AI cost comes from, and an honest read of what your FOCUS data shows about it.

You leave with

A shared AI cost map for your estate, its FOCUS coverage and gaps, and the set of questions your team can answer from the data it already has.

Is this for you?

  • AI spend is growing and the invoice explains nothing
  • Copilot seats and overage credits hit budgets nobody planned
  • Teams ship AI features without an owner for the bill

What we’ll work through

AI cost hides in platform features, assistant seats, prepurchase plans, and token usage. We build a shared map of where yours comes from, work out how much of it your cost data actually shows once it is in FOCUS (the FinOps Open Cost and Usage Specification, the standard shape billing data lands in), and identify the questions you can answer with what you have.

  • Where your AI cost comes from: model workloads, platform AI features, and assistants
  • Workload costs: Foundry, Fabric, Databricks, Azure ML, and supporting services such as AI Search
  • Assistant costs: GitHub Copilot, Copilot Cowork, Copilot Studio, and Security Copilot
  • How billing choices land in the data, including Agent Prepurchase Plans
  • What FOCUS tells you about GenAI models and token usage, and what it does not
  • The questions you can answer today, and the gaps that would need new data

What stays with your team

  • A shared AI cost map for your estate
  • Its FOCUS coverage, and a named list of the gaps
  • The cost questions your team can answer from the data it already has
  • The patterns that would close the gaps, if you decide to

What the price covers

Included

Your AI workload costs (Foundry, Fabric, Databricks, Azure ML, AI Search) and assistant costs (GitHub Copilot, Copilot Cowork, Copilot Studio, Security Copilot). How billing choices such as Agent Prepurchase Plans appear on the invoice and in FOCUS data. The signals FOCUS carries about GenAI models and token usage, and where the picture has gaps.

What we tailor to you

Which of your AI workloads and assistants we map, and which cost questions we work through against your data.

If you need more than this

Building a missing telemetry pipeline, or tailored allocation logic on top of the map, is a toolkit extension scoped separately.

Toolkit extensions

Related workshops