Skip to content

FREE CHECK

AI and Azure cost check

Test whether every AI and Azure line has a cap, an owner and an alert before the next invoice.

01Source02Boundary03Human decision04Evidence

WHAT TO REVIEW

Four questions for the team that owns the decision.

  1. Does every person who can use Cowork or an agent sit under a spending policy with a per-person limit and an alert?
  2. Is there a cap per agent on the environment where your teams build agents, with the owner on the alert?
  3. Do your Azure subscriptions carry budgets that alert, and token limits at the gateway for model deployments?
  4. Does someone read cost by use case against value each month and own the forecast?

Take the PDF with you.

Enter a work email, company and role to request the PDF. This action records an SIT inquiry for follow-up. Marketing emails require a separate choice.

We use these details to send the requested PDF and respond to this inquiry. Please do not enter customer records or credentials. Privacy policy Website terms

Download the PDF directly

THE CHECK

What does the AI and Azure cost check test?

Four things before the next invoice: whether every person who can use Cowork or an agent sits under a spending policy with a per-person limit and an alert; whether the environment where your teams build agents carries a cap per agent with the owner on the alert; whether your Azure subscriptions carry budgets that alert and token limits at the gateway; and whether someone reads cost by use case against value each month and owns the forecast.

Four questions, a PDF you can download directly with no email required, and the three fixes first.

THE MICROSOFT INSIGHT

Why is the spending policy the first question?

Because a spending policy is an access control as much as a budget: anyone in the scope of a policy that selects Cowork can use it whatever the limit, and a tenant-wide policy with a one-credit limit hands Cowork to the whole firm. The limit you set is the ceiling; the scope you set is the door.

Copilot Cowork is generally available, and a spending policy is its access control, not only a budget: any user in the scope of a spending policy that selects Cowork can use it whatever the credit limit, a tenant-wide policy with a limit of one credit grants the whole tenant, the old allow list under All Agents no longer decides access, and limit enforcement is asynchronous, so a user can start work after the limit is reached. (Microsoft Learn, checked Nov 18, 2026)

Set the scope and the limit together and the bill has a ceiling and a guest list. Set only the limit and the whole firm walks in under it.

THE METHOD

How does the check become a cost plan?

Each answer fixes one layer of the plan: the people, the agents, the platform, and the review.

  1. People: a default spending policy for everyone with an organization limit, a per-person limit and an alert threshold, then a policy per team with a model profile that keeps the frontier models out until the team needs them.
  2. Agents: a cap per agent on each Copilot Studio environment, with the agent owner on the alert.
  3. Platform: Azure budgets that alert, token limits per project and deployment at the AI gateway, and quota caps.
  4. Review: cost by use case against value each month, the forecast owned, and a power-user group with its own policy instead of raising everyone's limit.

THE BUDGET BLOCK

Can the platform stop an agent at a budget?

Not yet with a generally available control on Microsoft Foundry: the cost controls that alert or block a request at a budget are in preview, cost appears 8 to 24 hours after usage, and the request that crosses a threshold can still complete. So the plan caps tokens at the AI gateway, which is generally available, and treats the budget block as a preview to re-check.

Microsoft Foundry's cost and FinOps page lists what is generally available and what is preview: AI gateway token limits per project and model deployment, the Insights tab that finds recurring cost and token issues in an agent's traces, model router, prompt caching, batch and provisioned throughput, cost analysis by meter and tag in Azure Cost Management, Azure Cost Management budgets and quota caps are generally available; the Manage views for cost, fleet and operations, the per-project agent and model cost view, cost controls that alert or block requests at a budget, flex processing and Agent Optimizer are in preview. (Microsoft Learn, checked Nov 18, 2026)

Cap at the gateway and a looping agent stops at the limit. Wait for the block and it stops at the invoice.

See moreShow lessTwo more answers, then what will be working after two weeks and what this does not do yet.

THE DEFAULT

What changes on the Copilot invoice between January and March 2027?

Usage-based billing becomes the default for new Microsoft 365 Copilot Business licenses bought through CSP, with a default monthly spending limit for each user that an admin can change. The check asks whether that limit is one you chose.

Usage-based billing becomes the default for new Microsoft 365 Copilot Business licenses bought through CSP in Microsoft fiscal year 2027 quarter three, January to March 2027, moved from December 1, 2026 in the November 3, 2026 Partner Center update; the default spending limit stays at 4,000 Copilot Credits for each user each month and admins change it in Partner Center or the Microsoft 365 admin center. (Microsoft Learn, checked Nov 4, 2026)

Set the policies before the licenses land and the default is yours. Leave it and the default is the one you discover on the invoice.

THE NEXT RUNG

What comes after the check?

The two-week read-only Azure Cost Optimization and FinOps Assessment, scoped separately from any change and Microsoft-funded where you qualify; then the AI spend governance line of the Agent 365 engagement; then the monthly cost review inside managed AI and agent operations.

Each is its own scope, quoted after a scoping call. No percentage is promised and no amount appears on this site.

AFTER TWO WEEKS

What will be working after two weeks?

  1. You hold the PDF and the four layers are marked tested or assumed.
  2. Every person who can use Cowork or an agent is under a named spending policy, or the gap is the finding.
  3. The Azure budgets and the gateway token limits are listed, or their absence is the finding.
  4. You have booked, or decided not to book, the assessment.

HONEST ABOUT LIMITS

What this does not do yet

It does not make the Foundry budget block a control you can rely on today. Microsoft Foundry's cost and FinOps page lists what is generally available and what is preview: AI gateway token limits per project and model deployment, the Insights tab that finds recurring cost and token issues in an agent's traces, model router, prompt caching, batch and provisioned throughput, cost analysis by meter and tag in Azure Cost Management, Azure Cost Management budgets and quota caps are generally available; the Manage views for cost, fleet and operations, the per-project agent and model cost view, cost controls that alert or block requests at a budget, flex processing and Agent Optimizer are in preview. (Microsoft Learn, checked Nov 18, 2026) The cost controls are preview; the gateway token limit and the Azure budget are the controls the plan rests on.

It does not change a limit or a policy: the check reads, the assessment proposes, and you decide in writing.

It does not create a lead unless you ask for the PDF by email and say so.

THE ONE THING TO REMEMBER

Every AI and Azure line gets a cap, an owner and an alert before the invoice, not after it.

Said another way: the scope is the door, the limit is the ceiling, and the owner reads the forecast.

Said a third way: cap the tokens, not the ambition.

DIRECT ANSWERS

Questions before you decide.

Who should take this check?

The CFO or the budget owner and the person who runs IT, together. Answers that differ between them are the finding.

Is the PDF gated?

No. Download it directly. Asking for it by email is optional and creates an inquiry; marketing email is a separate box.

Does the check promise a saving?

No. No percentage and no amount appear on this site. The assessment models each change against your own baseline and shows the service risk beside the saving.

What is a spending policy?

A Microsoft 365 admin center control that grants selected agents and services to the users in its scope and sets Copilot Credit limits and alerts. It is an access control as well as a budget.

How is it priced?

The check is free. The assessment is quoted after a scoping call and scoped separately from applying any change.

NEXT STEP

Start with a boundary you can explain.

Choose a focused check, or speak with the team about the workflow you need to improve.