A version check can’t see order
In the demo above, a version-only check passes the rollback that ran its steps out of order. The log points at the current runbook version, but says nothing about the order the steps were followed in.
For platform, SRE, and operations teams
Ecdotic compares documented procedures with what actually executes, shows evidence behind any divergence, and drafts corrections for human review.
Demo
A demonstration of an Ecdotic run on synthetic data: a rollback runbook and sample deploy logs. The results, times and file hashes come from the run.
The problem
Runbooks get edited, workarounds become habit, and nobody reconciles the two until something goes wrong.
In the demo above, a version-only check passes the rollback that ran its steps out of order. The log points at the current runbook version, but says nothing about the order the steps were followed in.
Deploy, rollback and incident logs show what people actually did. Checking them against the runbook by hand is slow, and when it does happen, it’s usually after a critical failure. Why wait?
As more operational work runs through scripts and AI agents that read your docs, a stale procedure gets repeated instead of questioned, and the effects compound into organization-level failures.
How it works
01 Input
A runbook, plus the deploy, rollback or incident log for the times it ran. Exports are fine. Finding divergence only needs read access.
02 What Ecdotic finds
Ecdotic selects the procedure version in effect, matches each step to recorded events, and returns one of three results:
Matches Diverged Insufficient Evidence
03 What you get
Each result cites the runbook lines and log events behind it. For a divergence, Ecdotic drafts a correction and the owner decides whether it becomes the procedure.
Trust
Ecdotic observes what actually happened. Your team decides what becomes the procedure. How much Ecdotic does between those two is up to you.
An evaluation reads the procedure and the logs you share, and read-only remains a supported way to run Ecdotic. Further permissions only enable actions you authorize, such as publishing an approved correction.
Every result cites the exact runbook lines and log events it used, with a hash of each input file.
If the log is missing events, the result is insufficient evidence. A missing event is never treated as a skipped step.
A divergence shows that the runbook and the record disagree, not which one is right. The owner decides what becomes the procedure. Detecting a discrepancy never authorizes a change on its own.
Demos use synthetic data only. Before you share anything real, you get a one-page statement of what’s stored, where, for how long, and how it’s deleted.
This site sets no cookies and loads no third-party scripts, fonts or analytics. The booking calendar is Calendly’s, shown in its own frame.
If a correction is right, you shouldn’t have to copy it into documentation yourself. Ecdotic is built to carry a fix from evidence to a published procedure, with each step authorized by you.
Who’s behind it
A computer engineer from Georgia Tech with a background in AI, ML, and data engineering. I run every walkthrough myself. You’ll talk directly to the person building Ecdotic.
Evidence
Ecdotic is early. Here’s what’s been shown, what hasn’t, and what’s next.
FAQ
A procedure, such as a runbook in Markdown, and the execution record for the times it ran: deploy, rollback or incident logs. Exports are fine. Read-only access is enough for an evaluation and remains a supported way to run Ecdotic. Additional permissions only enable actions you authorize, such as publishing an approved correction to your runbook.
Only with your authorization. Today Ecdotic drafts the correction and your team applies it. We’re building approval-based updates: once an owner approves a specific change, Ecdotic publishes it or submits it through your existing review workflow, with the evidence and history attached. Letting Ecdotic apply defined categories of change within limits you set is something we’ll validate first.
No. It means the procedure and the record disagree. The runbook may be out of date, the execution may have departed from policy, or there may be an approved exception that isn’t written down. What happened isn’t automatically what should happen, so a person decides which side changes.
Ecdotic reports insufficient evidence and names the steps that have no matching events. It does not turn missing events into a finding.
A date tells you when a page changed, not whether anyone follows it. Ecdotic compares instruction steps with what the logs recorded, including their order.
The replay shows a real run of Ecdotic’s evaluator on synthetic data: a sample rollback runbook and three sample deploy logs. The desktop around it is illustrative; today Ecdotic runs as a command-line tool that writes a report.
Demos use synthetic data only. Before you share anything real, we agree in writing what is stored, where, for how long, and how it’s deleted.
Ecdotic is early. It runs today on synthetic data, and we’re looking for the first teams to evaluate it on one real procedure. Book a walkthrough to see whether yours is a fit.
Book a walkthrough
30 minutes with the founder. We’ll run the demo live, answer questions about access and review, and hear about a recent procedure that didn’t match what actually happened.
Calendar not loading? Book on Calendly.