Skip to content
SpendByTeam
Start free
NewCost per project for Claude Code

How much AI is each project really costing you?

Your AI coding bill arrives as one number. SpendByTeam breaks it down by team, project, branch and client — so you can budget it, defend it and bill it.

Start free — no card$npx spendbyteam init
  • Free for 3 developers
  • 2-minute setup
  • We never see code or prompts

Sample data. Teams, projects and clients are fictional.

Turn oneAI invoiceinto aper-projectledger yourCFOand yourclientsunderstand.The provider bill tells you the total. It doesn’t tell you which product, which client or which ticket the money went into. Now you’ll know.

How it works

Live in five minutes.

  1. 1

    Install

    Run one command

    Each developer runs the installer once — or an admin pushes one managed settings file to everyone. No proxy, no shared API keys, no change to how people work.

    npx spendbyteam init --token=sbt_ing_…
  2. 2

    Collect

    Usage flows in as numbers

    Your coding agent already emits OpenTelemetry metrics: cost, tokens, sessions, lines changed. We switch on metrics only. Logs and traces — the signals that could carry content — stay off.

    OTEL_LOGS_EXPORTER=none
  3. 3

    Attribute

    Every dollar lands on a project

    Repository → project → client, automatically. A tiny open-source hook adds branch and ticket, so you see cost per PR and per Jira ticket — not just per person.

    repo → project → client

It could be your dashboard.

By project, by team, by client, against budget. Switch the lens, keep the same numbers.

Projects

checkout-api

$3,184.52

41 PRs · $77.67 per PR

Teams

Platform team

$4,522.18

▲ 12% vs September

Clients

Kestrel Logistics

$4,148.82

2 projects · 3 repos

Budgets

ostara-patient-app

110%

Over budget by $241.08

All you need to keep AI spend under control.

Built for CTOs, engineering leads and agency owners who have to explain the AI bill.

Spend by project

Every repository rolls up into a project. Merge several repos into one product.

Team comparison

Compare teams side by side, month over month, with an end-of-month forecast.

Cost per PR & ticket

Branch and ticket IDs turn spend into cost per merged PR and per Jira ticket.

Budgets & alerts

Org, team, project or client limits. Slack and email at 80% and 100%.

Clients

Group projects under clients and see exactly what each customer’s work cost.

Client PDF reports

A monthly, client-ready PDF with ticket ranges and an optional handling fee.

Visibility controls

Open, admins-only or anonymous. Finance never sees per-person data.

No code, no prompts

Metrics only, attribute allowlist, open-source hook. Hosted in the EU.

Seat value

On seat plans, see the API-equivalent value each team gets from its seats.

For agencies & software houses

Stop eating your clients’ AI cost.

Group repositories under clients, set a handling fee, and send a monthly PDF that shows exactly what AI-assisted work cost on their projects — with ticket ranges, never prompts.

Example: 20-developer agency, ~70% of AI spend on client repos

$7,200 / per month recovered

Pass-through at cost + 15% handling fee. See the maths →

We count the spend.
We never see your code.

Content never leaves the developer’s machine: the exporters that could carry it are switched off, our endpoints discard anything that isn’t a metric, and what arrives is filtered through a fixed allowlist. The installer and hook are open source.

We receive

Cost, token counts, session ID, user email, repo URL, branch, ticket ID

Never

Prompts, responses, source code, file paths, shell commands, tool inputs

Read the security page
POST /v1/context — the entire hook payloadjson
{
  "session_id": "6f1c2a9e-4b7d-4e0a-9c55-0d2f8e1b7a43",
  "event": "SessionStart",
  "repo_url": "https://github.com/acme/checkout-api.git",
  "branch": "KST-482-rate-limits",
  "ticket": "KST-482",
  "at": "2026-10-17T09:41:12Z"
}

That’s it. Six fields, sent when a session starts or changes directory.

Pricing.

Per active developer, never a percentage of your AI bill. 14-day trial on paid plans, no card.

Compare all features →

Free

See the numbers before you talk about them.

$0

forever

Start free

Team

Budgets, forecasts and a monthly answer for finance.

$7

per active developer / month

Start 14-day trial

Most popular

Agency

Put the AI cost of client work on the client’s invoice.

$10

per active developer / month

Start 14-day trial

Business

Procurement, security review and invoice reconciliation.

Custom

annual contract

Talk to sales

Frequently asked questions.

Which coding agents do you support?

Claude Code today, via its built-in OpenTelemetry metrics. Other agents are on the roadmap.

Do you see our source code or prompts?

No. Logs and traces exporters are switched off, our ingest discards anything that isn’t a metric, and metric labels are filtered through a fixed allowlist. The full payload is on our security page.

How long does setup take?

About two minutes per developer with one npx command, or one managed settings file for the whole organization.

What counts as an active developer?

Anyone who sent at least one cost data point in the month. People who didn’t use an AI coding agent that month aren’t billed.

We pay per seat, not per token. Is this useful?

Yes. We show the API-equivalent value of usage, so you can see which teams and projects get the most out of the seats you already pay for.

Where is our data stored?

In the EU. Default retention is 13 months and is configurable per organization.

Next month, know where every AI dollar went.

Connect three developers free, forever. Add the rest of the team when the first report lands.

Start freeTalk to us