A cost with no owner does not get managed.
It gets discussed. Give the AI line an owner, a budget, a weekly number and a review, and it starts behaving like every other cost in the business.
Who should own the AI budget?
Ownership works best distributed to the teams that create the spend, with finance owning the consolidated position and the process. A single central owner cannot influence the decisions that drive cost, since those are made inside individual workflows. Camaze supports this by giving each team its own attributed number, budget, forecast and weekly digest, while finance holds the total, the allocation rules and the chargeback. Showback usually runs for a period or two before chargeback is switched on.
Everyone is concerned about it. Nobody is accountable for it.
Engineering assumes finance is watching the AI line. Finance assumes the team that built the system is watching it.
More detail
Both are reasonable assumptions and neither is true, so the number grows and the first person to genuinely examine it is whoever is assembling the board pack.
The underlying reason is structural. Provider bills are organized by API key, which maps to nothing anyone owns. Without attribution there is no owner, without an owner there is no budget, and without a budget there is nothing to be accountable to.
- Bills organized by API key rather than by team or product
- One shared key across several teams, so any split is a guess
- No per-team budget, so no threshold anyone can breach
- Spend surfacing only at close, when nothing can be changed
- Seats and tools assigned centrally with no cost back to the user
Illustrative product view. Figures are examples.
Attribution, then a budget, then a weekly number.
Ownership requires three things in sequence, and skipping any of them is why previous attempts have failed. A team cannot own a number it cannot see, cannot be accountable without a threshold, and will not stay engaged without a regular reminder.
- Every charge attributed to a team, project, workflow and product
- A budget per team with a named owner
- A weekly digest to that owner, in the channel they already read
- Alerts at thresholds, with the projected month-end effect
- Chargeback once showback has settled the disputes
The weekly number changes behavior before any money moves.
Showback consistently does the most work of any step, and it is the cheapest.
More detail
When a team sees its own AI spend, its trend and its position against an agreed budget, the conversation stops being about AI in general and becomes about a specific workflow that team controls.
It also surfaces the disputes early, while they are still cheap to resolve. Somebody will contest an allocation in the first month. That is the process working.
- A per-team view with budget, actual, forecast and variance
- Weekly digest to the owner in Slack, Teams or email
- Drill from a team total to the workflow and model driving it
- Largest month-on-month movements ranked, with causes attached
| Project | Model | Share | Cost | Change |
|---|---|---|---|---|
| Support automation | Claude Sonnet | $34,800 | +22% | |
| Search and ranking | GPT-4o mini | $21,400 | -6% | |
| Sales copilot | GPT-4o | $18,900 | +31% | |
| Docs assistant | Llama 3.1 70B, self-hosted | $15,200 | +9% | |
| Data enrichment | Gemini 1.5 Flash | $11,600 | -14% | |
| Internal agents | Mixed | $9,800 | +64% | |
| Engineering seats | Cursor, Copilot | $8,300 | 0% |
Illustrative product view. Figures are examples.
A threshold makes ownership real.
A budget is what converts a number into accountability. It does not have to be restrictive to work.
More detail
Most teams set budgets slightly above expected run rate, so the alert fires on genuine deviation rather than on normal growth.
Forecast breach alerts matter more than actual breach alerts, because they arrive with time left to respond. When the projection crosses the budget on day 12, the owner has 18 days.
- Budgets by team, project, workflow or provider
- Alerts on actual and on forecast, routed to the owner
- Projected month-end position included in every alert
- Budget performance tracked over time, per owner
Illustrative product view. Figures are examples.
Move it into the P&L that caused it.
Chargeback is the point at which AI enters the same discipline as every other cost.
More detail
It appears in a departmental P&L, gets defended in planning, and is owned by someone with a budget they care about.
Run showback for a period or two first. Chargeback introduced before the allocation is trusted generates arguments about the methodology rather than about the spend, which is the wrong conversation to have.
- Journal-ready export mapped to your chart of accounts
- Configurable split rules for shared platform and self-hosted costs
- Audit trail from a journal line back to the underlying usage
- Clean restatement when a mapping changes mid-period
| Report | Runs | Lands in |
|---|---|---|
| Monthly close pack | 1st of the month | Excel, emailed to finance |
| Chargeback journal | 1st of the month | ERP export, journal-ready |
| Board AI summary | Quarterly | PDF and slides |
| Cost by team | Every Monday | Google Sheets, live |
| Raw cost and usage | Nightly | Snowflake |
| Budget variance digest | Every Friday | Slack, #finance |
Illustrative product view. Figures are examples.
Give the line an owner.
Bring your team structure to a 30 minute session and we will show you your own spend attributed against it.
Rolling out ownership
The sequence that tends to hold, over about a quarter.
Attribute quietly
Connect, backfill and map to your org structure. Look at the result internally before publishing it. There will be surprises, and it is better to understand them before a team sees its number for the first time.
Publish showback
Give each owner their view and weekly digest. Expect allocation disputes in the first month and resolve them. This period is what earns the numbers their credibility.
Add budgets, then chargeback
Set budgets with owners once the figures are accepted. Switch on chargeback and journal export after that, when the methodology is no longer in question.
Keep reading
Cost allocation
Showback and chargeback by team, project, product and customer, from one source of truth.
Read moreFor CFOs and finance leaders
Explaining the AI line, attributing it and defending it in a budget review.
Read moreBoard reporting
The questions a board asks about AI spend, and the numbers that answer them.
Read moreQuestions people ask
Should one person own the whole AI budget?
Rarely. A single central owner cannot influence the decisions that actually create the cost, because those are made inside individual workflows by the teams that run them. Distributed ownership with finance holding the consolidated position and the process works considerably better.
What if teams dispute their allocation?
Expect it, and treat it as useful. Most disputes come from shared keys or shared platform costs, and both are worth resolving. Camaze shows the derivation of every allocated figure, so the argument is about the rule rather than about the arithmetic. Disputes almost always fall away after the first couple of periods.
Is chargeback necessary, or is showback enough?
Showback delivers most of the behavior change. Chargeback matters when AI spend is material enough to distort departmental P&Ls, when it belongs in cost of goods sold, or when you need it to appear in the planning cycle of the team creating it. Many companies stay on showback for a long time, which is a reasonable choice.
How do you handle shared platform costs?
With a split rule you define, applied consistently every period. Common bases are usage share, headcount or an agreed fixed allocation. Idle self-hosted capacity is usually held centrally rather than charged to whichever team used the endpoint last, which is both fairer and a better incentive.