Log reduction estimator

Measure the savings on your logs.

Chiron reduces your Splunk bill without losing any log data. The Chiron Estimator runs on a sample of your own logs to measure how much less you would need to index and how much you could save, before making any changes in production.

Repeated log lines combined into one Nine log lines on the left, six of them repeats. On the right, after Chiron, each group of repeats is one line and the three unique lines stay as they are, so there are five lines instead of nine. Your logs today Chiron With Chiron Log reductions
  • Runs on your machine, offline. It opens no network connection and sends nothing.
  • Your AI agent never sees raw logs. It only sees summaries and counts, with the details masked.
  • Measured, not modelled. The estimate comes from running Chiron's engine on your own logs.
  • You stay in control. Every setting is shown with its figures, and nothing is recorded until you approve it.
What you get

A number you can take to your team, and a plan to prove it.

A savings report

How much smaller your logs would get: an expected figure and a conservative one you can count on, broken down by step and by source. Add your daily volume and license rate and it shows the annual saving. Delivered as a PDF and as JSON.

Keep every line, or summarize

Repeated and similar lines are combined, and every line's content and fields can be rebuilt. Optionally, chosen sources become short summaries for a deeper cut. The report shows how much extra that saves and what it gives up.

A pilot configuration

The rules, collector transforms and Splunk settings a Chiron pilot needs to reproduce the estimate on your system, with a guide to where each file goes.

Works with the logs you have

Splunk exports, OTLP JSON, Kubernetes container logs and plain text files, compressed or not. No reformatting.

How it works

From download to report in about an hour.

How the estimator works A log sample goes into the estimator, which runs on your machine and produces a savings report and a pilot configuration. Log sample a folder on your disk Estimate offline, on your machine sends nothing Report detailed report and pilot settings

  1. Install it

    One command on macOS or Linux. It adds the estimator to Claude Code or Cursor, along with a small sample to try first.

  2. Point your agent at a log sample

    Export a few minutes from a busy cluster, or about an hour from a quieter one, and ask your agent for a savings estimate. A run takes roughly 15 minutes to an hour.

  3. Review and approve

    The agent walks you through each setting with figures the tool computed. You end with the report and the pilot configuration.

You need a Mac or Linux machine, Claude Code, Cursor or another MCP client, and free disk space of about three times your sample.