AI detection engine
Prometheus alert rules trigger within seconds of an anomaly. Metrics, logs and context enter one response path instead of six separate tools.
From four-hour on-call firefighting to 23-second AI detection.
67% incidents auto-resolved99.9% uptime maintainedMTTR cut to 14 minutes
ClientSaaS startup
Timeline4 weeks
RoleLead AI & DevOps Architect
StatusLive & deployed
01 / Evidence
See the system, not a mockup.
The screenshots below show the incident-response workflow, monitoring and operational outputs used by the system.
02 / The problem
Incidents were visible late, diagnosed manually and escalated inconsistently.
The operational cost was not only downtime. Engineers were losing sleep, context was scattered across tools, and each incident depended on whoever happened to be online.
→Engineers were being woken at 2 AM for alerts that often resolved on their own, high noise, weak filtering.
→Real incidents triggered manual diagnosis and copy-pasted logs in Slack, slowing down decisions under pressure.
→Critical incidents could sit unacknowledged for hours before the right person was escalated.
→Payment API failures could cascade before the team had a consistent automated detection path.
→Runbooks, post-incident context and pattern history were incomplete or scattered.
→Leadership escalation happened across ad-hoc channels without a consistent timestamped audit trail.
03 / The system
An incident pipeline designed around speed, evidence and escalation.
The system was built as a sequence of explicit responsibilities rather than one oversized “AI bot.” Each stage has a clear input, output and owner.
Prometheus
n8n
Claude AI
Slack
Jira
Notion / audit
Structured diagnosis
Claude AI turns raw incident context into probable cause, confidence level, immediate action steps and escalation recommendations.
Escalation ladder
Unacknowledged incidents move through a timed escalation path so reminders and leadership escalation are explicit and logged.
Remediation + audit trail
Eligible incidents can be remediated automatically while Jira, runbooks and reporting retain a complete history of what happened.
04 / The results
What changed after delivery.
23sMean detection timeDown from hours of manual monitoring
67%AI auto-resolutionIncidents closed without human action
14 minMean time to resolveDown from 4+ hour war rooms
99.9%AvailabilityMaintained across incidents
Stack used
05 / Client feedback
What the client said.
“Before SentinelAI, our on-call rotation was a nightmare. Engineers were getting paged at 3 AM for things that weren’t even real incidents. Since deployment, 67% of alerts resolve themselves before anyone wakes up. The AI diagnosis messages in Slack are so detailed that even our junior developers know exactly what to do. Bisma didn’t just build us a bot, she gave our team their nights back.”
JW
James WhitfieldCTO · FinTech Startup · United States
VerifiedHave a similar challenge?
Let’s talk about the failure mode, not the buzzwords.
Share the symptoms, current stack and operational stakes. I’ll tell you what I would investigate first.
Direct contact
bismaidrees@outlook.com
No pitch deck required. Start with what is breaking, what changed, and what is at risk.
Send project context ↗