Plutus

Blog

Notes on where the spend goes.

Your LLM gateway and your provider bill are both right

Put a gateway in front of your LLM providers and you get per-team attribution the invoice cannot give you. You also get two records of the same dollars, and adding them together is silent and expensive.

How to attribute AI API costs per key, model, and team

Your OpenAI bill is one number. Working out which key, which model, and which team produced it is a different job, and the two big providers make it possible in genuinely different ways.

How to split shared cloud costs between teams

Most of a cloud bill belongs to somebody. The rest, the shared cluster and the egress and the observability bill, is where chargeback quietly goes wrong. Here's how to divide it without the numbers stopping adding up.

From "spend is up" to root cause, without leaving the dashboard

A real anomaly from the Plutus demo account, and what it takes to go from a red number to knowing exactly why it happened.

Building a Plutus agent: MCP setup for Claude and Gemini

'Agent' just means something checks your spend on a schedule and tells you what it found. Here's the concrete setup in Claude and Gemini, not the pitch.

Root-causing a cost spike with the event overlay, start to finish

A walkthrough of what it actually looks like to go from 'costs are up' to 'here's why,' using the event timeline instead of a separate investigation.

The spike isn't the hard part. The investigation is.

A cost dashboard tells you the number moved. Finding out why is a separate afternoon. Here's how to make the chart answer the question itself.