I spent 6 years leading ops teams where nobody wrote documentation. So I built a tool that does it automatically.

For 6 years I ran 24/7 operations teams in iGaming and FinTech. One consistent pattern: incident postmortems almost never got written. Not because engineers didn't care — but because reconstructing a timeline, identifying root cause, creating Jira tickets, and formatting a Confluence page after a 3am incident is brutal. The knowledge disappeared every time.

I got laid off in January 2026 in a company-wide reduction. My manager explicitly told me performance wasn't the reason. Didn't make it easier.

Three months of job applications. Slow processes, ghosting, the usual. Then I decided to build the product I'd been wanting for 6 years.

What I built

Opsrift — 6 AI-powered tools for ops documentation and intelligence:

• Postmortem Generator (PagerDuty/OpsGenie/Datadog/Grafana → 9-section report → Jira push → Confluence publish)
• Shift Handover Generator (auto-fetch incidents from shift window)
• Runbook Generator
• Incident Assistant (paste or import alert → probable causes + investigation steps)
• Status Page Generator
• Incident Forecast (proactive risk signals from incident history patterns)

9 integrations: PagerDuty, OpsGenie, Jira, Confluence, Slack, GitHub, Datadog, Grafana, Microsoft Teams.

Stack: React/TypeScript, Express, Claude API, Clerk, Supabase, Paddle, Vercel. Built with Cursor. Zero traditional dev background.

The hardest part

Not the code. The positioning. First Reddit post got a lukewarm response because I led with "AI-powered" instead of the actual pain point. SREs are a skeptical audience — they hate hype and only trust tools they've seen solve a specific problem. Rewrote everything to lead with outcomes. Still applying that lesson.

Where I am now

Launched, live, working on distribution. Product is solid. Getting discovered is the real work.

Would love feedback from anyone in SRE, DevOps, or NOC — especially on Incident Forecast, which is the newest tool.

→ opsrift.com