Try a sample workspace. Trigger a spend spike, follow it into engineering output, and close the loop with an alert.
infercast / Acme workspaceInteractive demo · Sample data
Take it for a spin
A quiet week. Then a costly AI session. Can you spot what changed?
Your week in view
Jun 8–14 · 7 days
AI usage cost
$4,820
API-equivalent spend
Merged pull requests
100
From connected GitHub repos
Cost per pull request
$48.20
AI usage cost ÷ merged PRs
Active developers
6
Across 3 teams
Daily AI usage cost
USD
$540
Mon
$620
Tue
$680
Wed
$730
Thu
$850
Fri
$660
Sat
$740
Sun
Spend by team · 7 days
Backend$2,410
Frontend$1,446
Platform$964
A steady week across three teams. Simulate a spike to see how one session changes the picture.
Spend by AI tool
Same 7 days · USD
Claude Code
$1,700
Codex CLI
$1,610
Gemini CLI
$846
OpenCodeExperimental
$400
Grok Build
$264
The sample spike is a Codex session. Antigravity CLI & IDE are not connected in this demo; see integration status.
What did the spend support?
GitHub + Jira · 7 days
AI usage cost
$4,820
Same week, same teams
Merged pull requests
100
Delivery volume is unchanged
Cost per pull request
$48.20
Blended usage cost per PR
Review wait · P50
4.1h
Opened → first review
Delivery by initiative
Sample Jira rollup · linked through PR issue keys
ENG-24Checkout improvements
48 PRs
ENG-31Developer experience
32 PRs
Other workMaintenance & fixes
20 PRs
Signal Tower
Your team's attention queue
✓
No open alerts in this sample.
Simulate a spend spike to see a signal appear here.
Spend spike · BackendNeeds review
Friday spend more than doubled.
The simulated session added $1,000 in API-equivalent usage to the Backend team.
Before session
$850
After session
$1,850
Change
+118%
Acknowledging records that someone has reviewed the signal. It doesn't reduce spend or block AI requests.
Ready to explore. Run the sample scenario above.Local simulation · no account needed
Representative product views throughout this page. All numbers are sample data, not customer results.
Cost by engineer & AI tool
Different tools. One team budget.
Some engineers use Claude Code or Codex exclusively. Others mix Gemini CLI, OpenCode, and Grok Build into their workflow. See each person's tool mix and spend without stitching together separate reports.
Each bar uses the same dollar scale. Only tools with usage are listed. Total sample API-equivalent spend: $4,820.
01 / AI spend visibility
Know where every dollar goes.
One view of AI usage across developers, teams, tools, and models. Compare periods, investigate spikes, and take a clear cost breakdown into your next budget conversation.
Usage valued at API token rates is a comparison point; it is not an additional subscription charge.
02 / Engineering ROI & outcomes
Put the work next to the spend.
Connect GitHub to see merged PRs, cost per PR, review waits, and delivery trends. Give engineering and finance a shared view of what the investment supports.
Add Jira to roll up attributed AI spend and delivery by initiative. Trace work through issue keys in PR titles and branch names.
GitHub delivery metrics · Jira initiatives · Team comparisons
Engineering IntelligenceSample data · 30 days
Merged PRs
156
AI cost / PR
$80
Review wait · P50
4.1h
Jira initiative / ENG-24
Checkout improvements
Attributed AI cost
$3,840
Merged PRs
48
Issues resolved
18
Initiative costs are allocated from developer spend across linked PRs. Delivery metrics give context, not proof of causation.
03 / AI budget alerts & governance
Catch the exception. Keep the context.
Monitor budget thresholds, spending spikes, and inactive developers. Review the signal, acknowledge it, or snooze it while your team investigates.
Set policies for request cost, token usage, and allowed hours. Violations create monitoring signals and alerts; they do not block requests to AI providers.
Budget alerts · Policy monitoring · Audit logs
Signal TowerSample data
Budget threshold
80% of monthly budget used
$8,000 of $10,000 · Organization
Review spend and adjust your plan.
Policy violation
Request exceeded cost limit
$1.20 observed · $1.00 threshold
Recorded for review · Request not blocked
How it works
Your tools. A clearer picture.
Start with native OpenTelemetry. Add delivery context when you're ready. No application SDK changes for the guided CLI connectors.
01 / Create
Make it your workspace.
Sign up, name your organization, and create an API key for your team's telemetry.
02 / Connect
Configure your coding tools.
Use the guided setup for macOS, Linux, or Windows to connect Claude Code, Codex CLI, and Gemini CLI.
03 / Understand
Bring spend and work together.
See usage as telemetry arrives. On Team and Enterprise, connect GitHub repos and Jira projects for delivery context.
AI cost tracking for the tools your team uses.
Claude Code cost tracking
Follow Claude Code token usage, model costs, and developer activity through native OpenTelemetry. Compare API-equivalent usage with the Claude subscriptions you track to identify seats and model choices worth reviewing.
Codex CLI cost tracking
Bring OpenAI Codex CLI telemetry into the same cost dashboard as your other coding tools. Break down spend by developer and model, then use connected GitHub repositories to compare AI costs with merged pull requests.
Gemini CLI cost tracking
Track Google Gemini CLI usage through its OpenTelemetry exporter. Compare Gemini spend across engineers and teams, track subscription commitments, and monitor budget thresholds alongside Claude Code and Codex CLI.
Grok Build cost tracking
Connect xAI's Grok Build CLI through native OpenTelemetry logs. Track request tokens and reported costs alongside your team's other tools. Manual setup is available for its HTTP protobuf exporter.
OpenCode · experimental
Analyze individual model calls from OpenCode's experimental OpenTelemetry traces. Setup is opt-in: its upstream exporter can include prompt and response content, so review data sharing before enabling it.
Antigravity CLI & IDE
Google Antigravity CLI and Antigravity IDE are separate from Gemini CLI. Direct telemetry export to Infercast is not yet verified. We make that distinction explicit so you can plan around the integrations available today.
OpenTelemetry in. Clarity out.
Keep your existing coding workflow. Prompt capture is optional and off by default. Manage access with organization roles and export cost data as CSV.
All prices in USD. Choose the plan that fits your team.
Questions
Answers before you ask.
Setup · Costs · Delivery
Do you see our prompts or code?+
Prompt storage is optional and off by default. Guided setup explicitly disables prompt logging in Claude Code, Codex CLI, and Gemini CLI unless your organization opts in. OpenCode is an experimental opt-in connector whose upstream exporter can send content; review its data sharing before enabling it.
Do I need to install an SDK or change our code?+
No application SDK changes are needed. The guided setup configures the native OpenTelemetry exporters in Claude Code, Codex CLI, and Gemini CLI. Setup instructions are available for macOS, Linux, and Windows.
How do GitHub and Jira connect spend to output?+
GitHub supplies pull request, commit, and review data. Infercast combines this with AI telemetry to calculate metrics such as cost per merged PR. Jira issue keys in PR titles or branches link work to initiatives, with developer spend allocated across PRs. These metrics provide context, not proof that AI caused a delivery outcome.
Do policies block AI requests?+
No. Infercast monitors telemetry and records policy violations for review. Cost, token, and allowed-hours policies create alerts; they do not block upstream requests to AI providers.
How does pricing work?+
Free includes 5 developer seats and 30-day data retention. Team is $8 per seat per month for teams of up to 50 developers. Enterprise offers custom pricing and unlimited seats. No credit card is required to start free.
How do you track AI coding costs by engineer?+
Infercast collects usage telemetry from supported AI coding tools and associates it with developer identities. You can break down AI spend by engineer, team, tool, and model, compare periods, and review the subscriptions assigned to each developer.
How is AI cost per pull request calculated?+
AI cost per PR is the API-equivalent AI usage cost divided by the number of merged GitHub pull requests in the same reporting period. It is a blended team or developer metric. It does not measure the exact cost of one PR or prove that AI caused a productivity change.
Does Infercast support Antigravity CLI and Antigravity IDE?+
Direct Antigravity telemetry export to Infercast is not yet verified. Antigravity CLI and Antigravity IDE are separate Google products from Gemini CLI, whose native OpenTelemetry exporter has guided setup in Infercast.
Can I monitor AI budgets across multiple coding tools?+
Yes. Infercast brings Claude Code, Codex CLI, Gemini CLI, and Grok Build usage into a shared dashboard. OpenCode trace support is experimental. Budget threshold and spend spike alerts help teams review changes across tools. You configure the thresholds; Infercast monitors telemetry rather than blocking provider requests.
How are subscriptions different from usage costs?+
Subscription tracking records the monthly cost of tool seats you assign. API-equivalent usage values telemetry at model token rates so you can compare usage with that commitment. It is not an additional charge on top of your subscription bill.
Make the next budget conversation easier
Stop guessing at your AI bill.
Bring your team’s AI spend and engineering output into the same conversation. Start with 5 developers, free. No credit card required.