Track AI Cost per Department: See Which Teams Use AI, and What They Get for It
For CFOs and admins who need to tell each department head what their team's AI use costs, and keep each team inside its share.
To track AI cost per department, you need usage recorded against something you can group: a person, an app, an agent. A single provider invoice for the whole company tells you none of that.
Without it, AI spend becomes a shared overhead. Heavy users are subsidised, careful teams are penalised, and the budget conversation turns into guesswork.
This page covers what that costs, an illustrative example, and how Apragya's AI Credits reports support a department view.
What unallocated AI spend costs
Shared overhead
Nobody owns the bill, so nobody manages it.
Unfair split
Light users pay for heavy ones.
No return view
Spend cannot be set against a team's results.
Late surprises
The pool runs low before anyone notices.
AI cost allocation: how Apragya records usage
Apragya records every AI call with the user, the app, the model and the company, and reports agent, workflow and automation runs separately. There is no department field on usage, so a department view is built by grouping its people, apps or agents.
Provider invoices against Apragya usage reports
| Feature | Provider invoice | Apragya AI Credits reports |
|---|---|---|
| Per user | ✗ No | Top users and per-user allocations |
| Per app | ✗ No | Breakdown by app |
| Per agent | ✗ No | Usage by agent and workflow |
| Per company | ✗ No | Stamped on every call |
| Early warning | Spend cap | Alerts on plan share and unusual burn |
Checked against public information on 16 September 2026.
An example: a Noida electronics assembler
Take an illustrative 70-person company where sales, purchase and HR all use AI. The CFO wanted to show each head their team's share each month and keep HR's new resume screening within a set amount.
How the monthly department view was built
Users grouped
Each user listed under a department.
Usage by user
Credits per user for the month.
Usage by agent
Resume screening counted to HR.
Department totals
Summed per department.
Allocations set
Each person's share capped for next month.
The first report showed purchase using the most, mostly on reading supplier bills, which the head could justify. HR's screening agent stayed within its share because each recruiter's allocation was capped.
How to see AI spend by team
Step 1 of 5
Group people
Keep a list of users per department; teams can also be set up.
- Group people: Keep a list of users per department; teams can also be set up.
- Read usage: Use the breakdown by user, app, model and agent.
- One agent per purpose: Separate agents per department keep agent usage easy to assign.
- Allocate: Give each person an allocation per app; no allocation means no usage.
- Alert: Set alerts on plan share and unusual daily burn.
What sets Apragya apart for AI chargeback
One unit
Every AI feature is measured in AI Credits.
Per company
Groups can see usage by company, where a company is a unit.
Deny by default
Users without an allocation cannot spend.
Trends and forecast
Daily usage and a forecast for the period.
Alerts where you work
Email, Slack, Teams or webhook delivery.
Your own keys
Calls on your own keys are labelled by source.
Start with how AI Credits are metered, and see which agents drive usage, such as resume screening or budget overrun alerts.
Getting started
- List each department's users.
- Pull the per-user and per-agent usage for last month.
- Sum it by department and share it with each head.
- Agree each department's share for next month.
- Set allocations and an alert to match.
Showing the number is usually enough to change habits; caps are for the few teams that need them.
What it costs to start
AI cost by department questions
Is there a department field on usage?
Can we cap a department?
Can separate companies be separate cost centres?
Are agent runs included?
Can we export the numbers for our books?
What does it cost to start?
Give each department head their number
Group users and agents by department and share last month's AI Credits.