Watch every step stream
Tokens, timings, and outputs — live while the run executes. Every workflow run is auditable end to end, and usage is visible down to the credit.
An automation you can't inspect is an automation you can't trust. Arcflux streams each run as it happens and keeps the full step-by-step record after — what triggered it, what each step did, what it cost.
Live run streaming
- Watch steps execute in real time — LLM tokens as they generate, tool calls as they land.
- Timings per step; the whole run timeline at a glance.
- Test runs behave exactly like production runs — verify a flow before publishing.
The audit trail
- 100% of runs auditable, step by step: inputs, outputs, model, tokens, timings.
- Paused approval steps record who approved, when, and with what feedback.
- Failures show the failing step in context, not a stack trace in a log bucket.
Usage, down to the credit
- Credit usage per run — simple runs cost about 1 credit, multi-step runs with knowledge lookups 2 to 5.
- Workspace analytics: runs, usage, and activity across your workflows.
- Skipped events are visible too — see what the policy gate saved you.
run 4 steps · completed in 14s → triage summary + drafted Linear issue → four green steps on the execution timeline
Common questions
The short version of how observability works. Anything else: info@arcflux.ai.
Yes — credit usage is tracked per run, and a credit is roughly 1 cent of LLM usage.
Yes — steps stream live: tokens, timings, tool calls.
Yes — a test run executes the same steps with the same live streaming, so what you verify is what ships.
Yes — you can see what the policy gate dropped and why.
Turn your next event into a workflow
Model agentic workflows visually, trigger them from anywhere, and govern every run.