# BurnCap

> AI cost monitoring and budget guardrails for small teams. BurnCap ingests your LLM/API usage out-of-band (OpenAI or Anthropic admin-API import, CSV upload, or a lightweight SDK), attributes cost per day, model, feature, customer and environment, forecasts month-end spend, alerts on spikes and runaway agent loops, and exposes budget state through an opt-in `checkBudget` endpoint your app enforces — without ever proxying your model traffic or storing your prompts.

BurnCap is built for solo founders, AI "wrapper" products, agencies shipping AI tools, and product teams of 1–10 people who spend real money on LLMs but have no dedicated FinOps function. It is a budget-and-guardrail product, not a prompt/trace observability platform: it never sees prompts or completions, never proxies model calls, and its SDK defaults to fail-open budget checks.

- Official website: https://www.burncap.app
- One-line: Stop AI costs before they surprise you.

## Core use cases

- See exactly which AI feature, customer, model or environment is burning money.
- Forecast month-end LLM spend from your trailing usage before the invoice lands.
- Get alerted on cost spikes, runaway agent loops, expensive-model usage and dev/test spend.
- Set per-workspace / feature / customer / environment budgets with 50/80/100% thresholds and soft or hard caps your own app enforces via the SDK.
- Track per-customer unit economics and margin (cost vs. the revenue you record).
- Reconcile estimated spend against what OpenAI/Anthropic actually billed — every figure labeled estimated, provider-billed or forecasted.

## Key features

- Out-of-band ingestion: OpenAI & Anthropic daily admin-API imports, CSV import, and the `@burncap/sdk` events API with provider-response adapters (idempotent by request id — retries never double-count).
- Cost attribution by day, model, feature, customer_id and environment, with daily pre-aggregated buckets.
- Evidence-rated month-end forecast (with confidence, lag, and pricing-coverage limits) and a 0–100 "Burn Health Score" measuring whether you'd actually catch a cost surprise.
- Fresh budget state through `GET /api/v1/budget/check` plus optional atomic holds through `POST`; BurnCap never blocks traffic itself.
- Alerts via email (all plans) and Slack (paid), with a weekly or daily burn-report digest.
- Cost insights: cache-hit efficiency, what-if same-provider model swaps, and top movers.
- Team workspaces, project-scoped API keys, and ready-made AI-coding-agent setup prompts that use the SDK's provider-aware usage adapters.
- Privacy by design: no prompts/completions ever, provider credentials AES-256-GCM encrypted, API keys stored as hashes, one-click delete-all-data.

## Pricing

- Free — $0/mo: 10k events/month, 7-day history, 1 project, 1 seat, budgets & caps, email spike/runaway-agent alerts, Burn Health Score & forecast. No credit card.
- Starter — $19/mo: 100k events/month, 90-day history, 3 projects, 3 seats, daily provider auto-sync, Slack alerts, cost insights, weekly burn report.
- Growth — $39/mo: 1M events/month, 365-day history, 5 projects, 10 seats, per-customer profitability & margin alerts, daily burn report.
- Billed monthly through Lemon Squeezy (merchant of record); cancel anytime, data kept on revert to Free. Pricing page: https://www.burncap.app/pricing
- Eligible accounts receive 30 days of Starter automatically after signup/verification. No card or checkout, no automatic renewal or charge; access returns to Free unless the user chooses to upgrade. The offer has no published end date.

## Important pages

- Home: https://www.burncap.app/
- Pricing: https://www.burncap.app/pricing
- Agent setup prompts: https://www.burncap.app/prompts
- FAQ: https://www.burncap.app/faq
- Terms: https://www.burncap.app/terms
- Privacy (incl. cookies & subprocessors): https://www.burncap.app/privacy
- Refund policy: https://www.burncap.app/refund-policy

## Docs, SDK & support

- TypeScript/JavaScript SDK: https://www.npmjs.com/package/@burncap/sdk
- Support email: support@eastbase.studio

## Expanded machine-readable profile

- Full product profile (Markdown): https://www.burncap.app/llms-full.txt
- Structured discovery (JSON): https://www.burncap.app/ai-discovery.json