AI-Augmented IT Operations

Audit-ready by default

Graylog as the single observability sink. Automated cert renewal via ACME and EJBCA REST. NIS2 readiness PDFs generated on demand. AI-assisted triage of audit events. We turn certificate lifecycle from a fire-fighting exercise into a scheduled background job.

The problem

Most cert outages are observable hours before they bite

The signals are always there: a CT-log entry that doesn't match a known issuance, an OCSP response time creeping up, a renewal job that silently failed, a CRL that didn't refresh. The problem isn't visibility — it's that nobody is watching the right window.

Symptoms we see in cold engagements

  • Cert expiry surprises"Nobody told me api.example.com was expiring this weekend."
  • Mystery CT entriesA wildcard cert appears in the CT log and nobody can identify who issued it.
  • Stale CRLsThe CRL has been the same for a week — the publisher quietly stopped working.
  • NIS2 panicAudit deadline in six weeks, no inventory, no evidence trail.

What we put in place

  • Active monitoringExternal Cert Monitor SaaS for the perimeter, internal probes for private services.
  • Renewal automationACME for the easy 80%, REST automation for service-account mTLS, escalation paths for the rest.
  • CT-log diffingA daily diff against a known-good cert inventory. Unknown issuance triggers an alert.
  • NIS2 evidence packA quarterly PDF with the inventory, controls map and incident log — ready for the auditor.
The stack

One sink. Many feeds.

A simple architectural rule: every event in your operation lands in Graylog, in a structured form, with consistent labels. Once that's true, queries are cheap and dashboards write themselves.

📊

Graylog GELF sink

Graylog OpenGELF UDP/TCPstreams

Every component emits structured GELF. Streams route by tag, pipelines extract canonical fields, dashboards are built per audience: ops, security, compliance.

🔄

Renewal automation

ACMEEJBCA RESTSalt

ACME for the web tier, REST for service mTLS, Salt distributes the renewed bundle. Failures emit GELF alerts; success goes silently.

📡

CT-log monitor

crt.shCT pollingdiff alerts

Polls CT logs for your domains and diffs against a known inventory. Unknown issuance is the canary for credential compromise or shadow IT.

📋

NIS2 evidence

NIS2 Art. 21control mappingPDF report

A quarterly PDF: cert inventory, control mapping, incident log, key-ceremony attestations. Ready to hand to the auditor.

🤖

AI-assisted triage

Claude APIstructured outputrunbook gen

LLM-assisted: summarise audit feeds, draft incident timelines from the GELF stream, generate first-pass runbook updates from change tickets.

📈

Dashboards & SLOs

Graylog dashboardsPKI Ops Dashboard

Ops dashboard: total certs, expiring, critical, CRL freshness. Security dashboard: failed handshakes, unknown CT entries, revocation events.

Outcomes

What changes after deployment

  • Zero-touch renewalsThe 80% of certs that can renew via ACME do — without anyone being paged.
  • Median expiry alert lead time: 30 daysNo more 02:00 "the cert expired" messages.
  • CT-log coverage for every owned domainA daily diff. Unknown issuance triggers an investigation.
  • NIS2 audit evidence on demandGenerate the PDF report yourself, anytime.
  • One query, one timelineGraylog answers "what happened to that cert?" in seconds, not hours.
  • Runbooks that stay currentLLM-assisted updates after every incident or change.

Need a NIS2 readiness assessment?

Free 60-minute discovery call. We map your current cert inventory, observability and audit posture against NIS2 Article 21 and tell you what to fix first.