Granular Executions
Understand the cause, not just the effect, by identifying the executions that drive cost and value
Explore incoming live traces
Watch executions flow through your AI system in real time. Each trace represents a complete unit of work: an agent run, a retrieval chain, a multi-step workflow, captured as it happens.
Filter and sort the live stream by cost, latency, status, or any custom dimension to surface what matters most before it compounds.
Cost Over Time
Spike Propagation
Understand detailed trace costs & value
Open any trace to break down exactly where money was spent and what value was delivered. See every step in the execution path (each model call, each tool invocation, each external request) with cost and value attributed individually.
Compare traces side by side to understand why two similar workflows produced different cost profiles. Was it a longer reasoning chain? A cache miss? A retry loop? The trace tells you.
Run queries on traces
Go beyond browsing. Write queries to search, filter, and aggregate across your full trace history. Find every trace where cost exceeded a threshold, where a specific service took longer than expected, or where value per execution dropped below baseline.
Save queries as reusable components to track patterns over time. Pipe results into alerts and dashboards so the right team gets notified before a trend becomes a problem, or use them as building blocks in your economic graphs.
FREE for developers, forever.
Signup and connect your telemetry and billing pipelines to start tracking unit economics across your AI systems in minutes.