Scan. Sift. Verify. Repeat.
In 2019 I was in the trenches — working as a penetration tester and cybersecurity engineer, splitting my time between pentests, incident response, and building the tooling that held it all together. The work had a rhythm, and too much of that rhythm was rework: re-running port scans, hand-checking service versions, reconciling results that had shifted since the week before.
And on large projects, things got missed. Not through carelessness — the team was good, and everyone was looking. There was simply more surface than any of us could hold in our heads at once.
Teach the machine to read the scans.
So we started digging into machine learning — models that could create, analyze, and act on scan data instead of waiting for a human to get around to it. The rework shrank. The misses shrank.
Through 2020 and 2021 the concept became a build: piece by piece it took on more of the testing work itself, from analyzing service versions to acting on what it found. Somewhere in there it stopped being a helper and started being the searcher.
Incident response showed the other half.
Incident response showed me where breaches actually take hold: in the gaps. The organizations I walked into had tools — often excellent ones — and capable teams. But those capabilities were spread across many teams and consoles, and in the spaces between them, things got missed.
It was the same pattern from the scan work, at enterprise scale: nobody misses on purpose. There’s just more surface than attention.
Proof has a shelf life. Nobody was checking the date.
Build the thing that closes it.
The build kept evolving until it became Rimator: a standing loop that carries the testing work end to end, then plugs into the controls you already own — EDR, SIEM, cloud, identity — to validate that every attack was actually seen.
Seven years from a frustrated port scan to going to market in 2026. Not another tool on the pile — the thing that tests the pile, continuously, kept honest by its own evidence.
— Bogdan Salamakha · Founder, Rimator