The detection logic is fully transparent and editable, and that's the thing I'd point to first. Every rule is human-readable, so when we see a new phishing pattern we can fork an existing rule, tune the conditions, and have it live the same day instead of filing a vendor ticket and waiting a release cycle. The rules live as code, which means we version control them and code review them like any other engineering work. Onboarding was genuinely fast. It connects to our mail tenant over API rather than sitting inline, so there were no MX changes, no mail flow risk, and no added delivery latency. We were reviewing real detections within days of signing, and we could run in observation mode first to see what it would have caught before letting it take action on anything. The verdict detail is what changed our day-to-day workflow the most. Instead of a black-box risk score, we get the specific signals behind a detection along with full message and attachment analysis, so triage on a reported message takes a couple of minutes rather than a manual header-and-link investigation. Being able to query historical mail with a real query language means email is now something we can actually hunt across, which we didn't have before. Integrations have held up well. The API is complete enough that we pipe detections into our own SIEM and automation rather than living in one more console. Detection quality on business email compromise, vendor and executive impersonation, credential phishing, and QR-code lures has been consistently strong. Support is responsive and technically credible, with direct access to people who know the product instead of a tiered queue.

I love the idea of crowdsourcing detection rules. You can take a new, novel phishing campaign and quickly turn it into a detection rule, backcheck it, and share it with the community, creating a kind of herd immunity, in a matter of minutes.

Excellent tool, efficacy right on par with Abnormal. More functionality from administrative perspective.