Shared endpoints
Cost allocation mapping
A mapping of one shared model endpoint or GPU pool onto the product teams that actually call it, so the bill is no longer a single unexplained number.
Who it is for
Companies where two or more products share an endpoint, a reserved GPU pool, or a single vendor contract, and finance wants a split that the teams will recognise.
What you leave with
An allocation table with the method written in plain language — by token count, by call volume, by reserved hours, or a hybrid you agree in the room — plus the caveats of that method.
Scope
One shared resource and up to four calling teams. More teams are quoted.
- Duration
- Two to four weeks, depending on how many teams must be interviewed.
- Where
- Kuala Lumpur or video. Interviews are scheduled around the teams, not around us.
- Fee basis
- From RM 9,600. Each additional team beyond four is quoted.
- Who leads
- Lim Wei Han leads. Farah reviews the method before it leaves the office.
Included
- Interviews with each calling team (thirty minutes each)
- A draft method, a challenge session, and a final table
- A one-page note on what the method cannot see
Not included
- Building chargeback software
- Changing how teams call the model
How the sitting runs
-
Inventory
We list who calls the shared resource and what evidence exists (logs, tags, or none).
-
Method
We propose a split and sit with the teams until the objections are on the page.
-
Table
Finance receives the table and the caveat note.
Preparation
A list of calling teams, a billing owner, and whatever tags or request IDs you already have. Missing tags are common; we work with call volume if that is all that exists.
Limits we keep
An allocation is a method, not a truth. If two teams refuse a method, we document the disagreement rather than invent a split they will not use.
Next step
Name the shared endpoint and the teams. We will say whether the evidence is enough to map.