Project Ledger
Tracks usage and cost across AI providers and internal applications, identifies optimization opportunities, and provides budgets and alerts so spend stays predictable as usage scales.
Product status
- Concept
- Prototype
- MVP
- Pilot
- Beta
- Production
- Scale
Problem
Teams building on AI APIs frequently lack visibility into what's driving spend, and discover cost overruns after the fact.
Why this problem matters
AI API costs can scale unpredictably with usage; without attribution and budgeting, spend becomes a recurring finance and engineering friction point.
Solution
A usage-tracking layer that attributes spend to application and user, surfaces optimization opportunities, and enforces configurable budgets with alerts.
Core capabilities
- Multi-provider usage tracking
- Budget alerts
- Cost attribution by application
- Optimization recommendations
Architecture
A lightweight tracking layer captures usage per request, attributes it to application-level metadata, and stores aggregated usage for budgeting, alerting and optimization recommendations.
How it works
- 1Route API calls through the tracking layer.
- 2Usage is attributed to application and user/team.
- 3Dashboards surface spend trends and cost drivers.
- 4Configurable budgets trigger alerts before overruns occur.
AI capabilities
- Usage attribution
- Cost analysis
- Optimization recommendations
Security model
- • API keys handled via environment-scoped secrets, never logged in plaintext
- • Tenant-isolated usage data
Not yet independently assessed
Integrations
- • OpenAI-compatible APIs
- • Anthropic API
- • Planned: additional model providers
Deployment models
Target customers
- • Engineering teams building on LLM APIs
- • Platform and FinOps teams
- • AI product teams
Business model
Usage-based SaaS pricing tied to tracked token volume, with a free tier for early adopters.
Differentiators
- • Attribution down to a granular usage unit, not just provider-level billing totals.
Evidence
Product evidence
Live sandbox
Available on requestEngineering evidence
Repository
Available on requestTraction
Stage
Beta
Production pilots
Data available upon diligence
Roadmap
now
Multi-provider tracking
Stable attribution across major LLM providers.
next
Optimization engine
Automated caching and right-sizing recommendations.
later
Budget automation
Automatic throttling or routing when budgets are at risk.
Team capability
Investor relevance
The most mature product in the portfolio today, demonstrating the company's ability to ship and iterate a working SaaS product end to end.