Catch the spike before the invoice
AI cost monitoring for small teams: attribute spend by feature, customer and model, forecast month-end, enforce budgets through opt-in checks — no proxy, no stored prompts.
Free plan, with Starter and Growth tiers at $19 and $39 a month.
The problem
AI products start cheap and turn unpredictable in production. Provider dashboards report spend only after the money is gone, and none of them tell you which feature or customer caused it.
The approach
BurnCap ingests usage out-of-band — provider imports, a CSV, or an SDK — and attributes every dollar to a feature, customer and model. It forecasts month-end spend, alerts on spikes and runaway loops, and lets your own app enforce budgets through a check that fails open.
What's inside
Cost by feature, customer and model
Month-end spend forecast
Spike and runaway-loop alerts
Budgets with soft and hard caps
OpenAI, Anthropic and CSV imports
TypeScript SDK and budget-check API
Unit economics and margin alerts
Cost insights and what-if model swaps
04From the lab
The decisions, trade-offs, and engineering notes behind BurnCap — written as we build.
AI products are cheap to prototype and unpredictable in production. Provider dashboards log the burn — none of them stop it. The first note in an open build log.
BurnCap's Insights feature is the smart part of an AI cost tool — and it calls zero models. Why the most AI-native decision we made was choosing not to use AI.
How BurnCap keeps its own API costs and latency bounded with three cheap layers — and a Redis layer built so that losing Redis changes nothing.
Integration code is written by agents now, not by hand — so BurnCap ships the prompt, not a quickstart. Why a prompt can be an executable spec, not just copy-paste.
More from Eastbase
