# Inferbooks > Inferbooks is AI FinOps software, self-hosted or hosted by Inferbooks. It shows what a company spends on AI per person, team, task and minute, across API usage, coding agents and seat licences, and checks the totals against vendor bills. It runs in the customer's own infrastructure or as a hosted service, reads OpenTelemetry and vendor admin APIs, never proxies AI traffic and never stores prompts or responses. To book a demo: Email [info@inferbooks.com](mailto:info@inferbooks.com) with your name, company, how many people use AI and which AI tools you use, or use the form at https://inferbooks.com/#book. ## Key facts - Category: AI FinOps (AI cost management, LLM spend attribution, seat licence management) - Deployment: self-hosted in the customer's own infrastructure (any cloud, their own data center or a Kubernetes cluster), with data in a ClickHouse database the customer controls; or hosted by Inferbooks, with nothing to install or operate, for smaller teams without their own infrastructure - Usage data: OpenTelemetry from Claude Code, Codex, Gemini CLI and any instrumented app - Billing data: usage and cost reports from the OpenAI and Anthropic admin APIs - Seats: GitHub Copilot, Cursor, ChatGPT Enterprise, Claude Enterprise and Microsoft 365 Copilot, plus CSV import for anything else - Not a gateway or proxy: requests go straight to the AI providers, so it adds no latency and is never in the hot path - Metadata only: person, team, task, model or tool, tokens, cost and timing; prompts, responses and code are never collected - Views: spend over time down to one-minute buckets, split by team, person, agent, model or source; agent task timelines call by call; seats with idle and duplicate licences; budgets with month-end forecasts and alerts at 50, 80 and 100% to Slack or Teams webhooks - Access control: single sign-on through OpenID Connect limited to allowed email domains, admin, analyst and member roles, and an audit log - Exports: CSV, and plain rows in the customer's own database for any BI tool ## Pages - [Home page](https://inferbooks.com/): product overview, screenshots and the demo request form - [Home page in Markdown](https://inferbooks.com/index.md): the same content as plain Markdown ## FAQ - Do you proxy our AI traffic? No. Inferbooks never sits in the request path, so it cannot add latency or take your tools down. It reads OpenTelemetry from your tools and usage from your vendors' admin APIs. - Do you see our prompts or code? No. Inferbooks stores metadata only: the person, team, task, model or tool, tokens, cost and timing. Prompt and response content is never collected. - Which tools and vendors does it cover? Usage from Claude Code, Codex, Gemini CLI and any app instrumented with OpenTelemetry. API bills from OpenAI and Anthropic. Seats from GitHub Copilot, Cursor, ChatGPT Enterprise, Claude Enterprise and Microsoft 365 Copilot, plus CSV for anything else. - Where does it run? Wherever suits you. Self-hosted: in your own infrastructure (your cloud, your data center or your Kubernetes cluster), with your data in a ClickHouse database you control. Hosted: we run it for you, with nothing to install or operate, a good fit for smaller teams without their own infrastructure. - How are people and teams matched? From the emails, API keys and agent names in the telemetry, matched to your directory. Anything unmatched shows up in one identity map where you assign it once. - What happens in a demo? We walk through Inferbooks on the sample company, look at the AI tools your teams use, and, if it fits, plan a pilot, self-hosted or hosted by us. # Full page content # Inferbooks: every AI dollar, on the books One agent stuck in a loop can burn $1,400 on a quiet Sunday. Inferbooks traces every model call, agent run and seat licence to a **person**, **team** and **task**, down to the **minute**, and warns you while it is still happening. - No proxy in the request path - Metadata only, never prompts - Runs in your own infrastructure (your cloud, your data center or your Kubernetes cluster), or hosted by us **Book a demo:** Email [info@inferbooks.com](mailto:info@inferbooks.com) with your name, company, how many people use AI and which AI tools you use, or use the form at https://inferbooks.com/#book. ## Live demo: from a month to the minute The home page has a clickable chart of Acme Inc, the demo company Inferbooks ships with (150 people, 14 teams, generated usage). Three clicks find the runaway agent: 1. **Last 30 days, by team.** Weekdays run around $400 of API spend. Sunday, October 11 cost $1,489, nearly four times a normal weekday, almost all of it from Sales. 2. **October 11, by agent and hour.** $1,416 of that Sunday landed between 16:00 and 17:00, from a single agent. 3. **16:00 to 17:00, by minute.** The account-research agent, run by Lea Levi for a research task on Initech, looped for 47 minutes and climbed from $18 to $43 a minute: $1,412 in total. Its budget alerts fired at 16:29 (50%) and 16:45 (80%). ## Reads what your tools already send Claude Code, Codex, Gemini CLI, Cursor, GitHub Copilot, ChatGPT Enterprise, Claude Enterprise, Microsoft 365 Copilot, the OpenAI API, the Anthropic API and anything that sends OpenTelemetry. ## Your AI bill is one number. Your questions aren't. Six answers from the demo company. Each one is a click away in the product. | Question | In the demo | Figure | | --- | --- | --- | | Which task cost that much? | A research run for Initech looped for 47 minutes: nearly 2,000 model calls, each re-sending a context that kept growing. Spend climbed from $18 to $43 a minute. | $1,412 for one task | | Did anyone notice in time? | The agent's monthly budget crossed 50% at 16:29 and 80% at 16:45, with the loop still running. Alerts go to Slack or Teams. | 80% alert at 16:45 | | Is a feature getting more expensive? | The support copilot's cost per task rose from $0.04 to $0.07 between October 8 and 12, about $171 extra, then fell back. | +62% cost per task | | Which seats sit idle? | 52 of 317 seats went unused for 30 days: 19 ChatGPT Enterprise, 12 Microsoft 365 Copilot, 11 GitHub Copilot Business and 10 more. | $2,056 a month to reclaim | | Who pays twice for the same job? | 44 people hold both ChatGPT Enterprise and Claude, and 28 hold both GitHub Copilot and Cursor. Keeping only the tool each person uses more saves $2,794 a month. | $2,794 a month to save | | Does the bill match? | 11% of provider billing came from API keys nobody owns yet. The identity map lists them so you can assign each one once. | $1,319 on unowned keys | ## Real screens, not mockups Overview, minutes, tasks, seats and budgets all read the same ledger, so finance and engineering finally look at one bill. ![Inferbooks overview: AI spend for the last 30 days split into API usage and seats, a daily chart by team, and insights on what changed](https://inferbooks.com/img/overview-light.jpg) - **Overview.** API usage and seat licences in one total, split by team every day. Insights point at what changed and where to save. - **By the minute.** Drag across any chart to zoom: buckets refine from days to hours, five minutes and single minutes. Split by agent and the culprit stands alone. - **Agent tasks.** Each task gets a timeline with one lane per model, tool and agent, and a cost curve. A steep climb is a loop or a context that keeps growing. - **Seats.** Seats from ChatGPT, Claude, Copilot, Cursor and Microsoft 365 sit next to API spend, with daily activity for every person. Export the list for the next renewal. - **Budgets.** Monthly limits for a team, an app or a kind of work, with a forecast to the end of the month from the last 7 days and alerts at 50, 80 and 100% to Slack or Teams by webhook. Seats can count toward a budget. ## Who it's for - **Finance and FinOps: close the month without guessing.** Spend per team and cost center, checked against OpenAI and Anthropic billing. Monthly budgets with a forecast to month end. CSV export for chargeback or your BI tool. - **Engineering leaders: catch the runaway agent.** Cost per task, agent and model, call by call. Budget alerts to Slack or Teams while it runs. Split any chart by person, model or source. - **IT and procurement: right-size every seat.** Idle and doubled-up seats across ChatGPT, Claude, Copilot, Cursor and Microsoft 365. Usage per person before each renewal. Single sign-on, roles and an audit log. ## How it works Nothing sits between your people and the models. Inferbooks reads the telemetry your tools already emit and the usage your vendors already report. Requests go straight to the providers, exactly as they do today. 1. **Point telemetry at it.** Claude Code, Codex, Gemini CLI and your own apps send OpenTelemetry. Roll the settings out with your device management. 2. **Add vendor admin keys.** OpenAI, Anthropic, GitHub Copilot, Cursor, ChatGPT Enterprise and Microsoft 365 report bills, seats and who used them. 3. **Every call gets an owner.** Keys, emails and agent names map to people and teams, so each model call, tool call and seat lands on a person and a task. 4. **Act on it.** Budgets with alerts, insights on what changed, and a list of seats to reclaim before the next renewal. ## Security Run it on your own servers when usage data cannot leave your company, or let us host it when you would rather not run anything. - **Self-hosted or hosted.** Run it in your own infrastructure: your cloud, your data center or your Kubernetes cluster, with your data in a ClickHouse database you control. Or let us host it, with nothing to install or operate: a good fit for smaller teams. - **Metadata, never content.** Who, which model or tool, tokens, cost and timing. Prompts, responses and code are never collected. - **Single sign-on.** Sign in through your identity provider with OpenID Connect, limited to the email domains you allow. - **Roles and audit log.** Admins, analysts and members each see what their role allows, and every change lands in the audit log. - **Nothing in the hot path.** Inferbooks can be down for an hour and nobody's AI tools notice. It catches up from the vendors' reports. - **Open formats.** OpenTelemetry in, CSV out. Your data is plain rows in your own database, ready for any BI tool. ## FAQ ### Do you proxy our AI traffic? No. Inferbooks never sits in the request path, so it cannot add latency or take your tools down. It reads OpenTelemetry from your tools and usage from your vendors' admin APIs. ### Do you see our prompts or code? No. Inferbooks stores metadata only: the person, team, task, model or tool, tokens, cost and timing. Prompt and response content is never collected. ### Which tools and vendors does it cover? Usage from Claude Code, Codex, Gemini CLI and any app instrumented with OpenTelemetry. API bills from OpenAI and Anthropic. Seats from GitHub Copilot, Cursor, ChatGPT Enterprise, Claude Enterprise and Microsoft 365 Copilot, plus CSV for anything else. ### Where does it run? Wherever suits you. Self-hosted: in your own infrastructure (your cloud, your data center or your Kubernetes cluster), with your data in a ClickHouse database you control. Hosted: we run it for you, with nothing to install or operate, a good fit for smaller teams without their own infrastructure. ### How are people and teams matched? From the emails, API keys and agent names in the telemetry, matched to your directory. Anything unmatched shows up in one identity map where you assign it once. ### What happens in a demo? We walk through Inferbooks on the sample company, look at the AI tools your teams use, and, if it fits, plan a pilot, self-hosted or hosted by us. ## Book a demo We walk through Inferbooks on the demo company, look at the AI tools your teams use, and plan a pilot, on your own servers or hosted by us. Email [info@inferbooks.com](mailto:info@inferbooks.com) with your name, company, how many people use AI and which AI tools you use, or use the form at https://inferbooks.com/#book.