BriefOps AI

BriefOps AI transforms production alerts into coordinated incident response workflows with AI-generated briefs, war rooms, runbooks, communications, and postmortems.

Comment

Inspiration

Production incidents are chaotic. Alerts arrive from tools like Alertmanager, Prometheus, Grafana, Datadog, logs, and webhooks, but responders still have to manually triage the alert, understand customer impact, write incident summaries, create a war room, draft stakeholder updates, open issue reports, and prepare postmortems.

I built BriefOps AI to turn that messy alert-to-response process into a structured workflow. The goal is simple: when a production alert fires, the team should immediately get an AI-generated incident brief, a response workspace, a runbook, tasks, communication drafts, timeline events, and auditable artifacts.

What it does

BriefOps AI is an AI-powered incident response workspace for DevOps, SRE, and platform teams.

The core workflow is:

  1. A production alert is received from Alertmanager or another monitoring source.
  2. BriefOps AI stores it as an Incoming Signal.
  3. AI analyzes the signal and generates a structured incident brief.
  4. The user creates a War Room from the brief.
  5. The War Room provides:
    • Overview
    • Runbook
    • Tasks
    • Communications
    • Timeline
    • Postmortem
  6. The team can send or copy updates to Slack, Discord, Telegram, or GitHub Issues.
  7. Every copied, sent, downloaded, or exported output becomes an Incident Artifact for auditability.

BriefOps AI helps teams move from alert noise to coordinated incident response.

How I built it

I built BriefOps AI with MeDo as a full-stack production-style SaaS application.

I structured the build as a series of vertical slices instead of one large prompt:

  1. Authentication and team onboarding
  2. Connection management
  3. Alertmanager webhook ingestion
  4. AI-powered incident analysis
  5. Incident Inbox workflow
  6. War Room workspace
  7. Runbook and task synchronization
  8. Stakeholder communications
  9. Timeline and incident events
  10. Incident Artifacts
  11. Outbound delivery to Slack, Discord, Telegram, and GitHub
  12. Landing page and product polish

The backend uses Supabase Auth, Supabase Database, Row Level Security, and Supabase Edge Functions. Webhook ingestion is handled through Edge Functions, while AI analysis generates structured incident packages that power the War Room experience.

The most impressive feature

The most impressive feature is the end-to-end incident workflow:

Alertmanager → Incoming Signal → AI Incident Brief → War Room → Communications → Incident Artifacts

A real Alertmanager-compatible webhook can create a signal inside BriefOps AI. From there, the AI analysis creates a structured incident brief with severity, confidence, customer impact, root cause hypothesis, recommended runbook, action items, stakeholder messages, and postmortem draft.

The War Room then turns that analysis into execution: responders can follow runbook steps, create linked tasks, track timeline events, tune communications, send updates, create issue reports, and preserve all outbound outputs as Incident Artifacts.

Challenges I faced

One major challenge was making the system feel like a real production workflow instead of a static AI demo. I had to carefully separate:

  • Incoming Signals as raw alert data
  • Incidents as AI-generated analysis
  • War Rooms as operational response state
  • Incident Events as timeline/audit history
  • Incident Artifacts as copied, sent, downloaded, or exported outputs

Another challenge was synchronization. When a responder changes the incident phase, completes a runbook step, creates a task, or sends a message, the War Room needs to update the timeline and related views without overwriting manually edited communication drafts.

I also had to handle real-world integration problems such as webhook token validation, connection status, failed outbound delivery, retry behavior, and preventing AI analysis jobs from getting stuck.

What I learned

I learned how to use MeDo more effectively by prompting in smaller production-focused slices. Instead of asking for a full app at once, I got better results by describing one workflow at a time, defining the data model, the expected UI behavior, and acceptance tests.

I also learned that AI apps need strong state models. A good AI-generated summary is useful, but a production system also needs persistence, event history, retries, permissions, and clear user workflows.

What's next

Next, I would improve BriefOps AI with:

  • More production monitoring integrations
  • Better alert grouping and correlation
  • Team-specific AI analysis templates
  • Real-time collaboration inside War Rooms
  • Automated incident severity scoring
  • Postmortem export to Markdown, GitHub, and documentation tools
  • More advanced delivery retries and audit controls

BriefOps AI is designed to become an incident response operating system for production teams.

Built With

  • alertmanager-webhooks
  • discord-webhooks
  • github
  • issues
  • large-language-model-capability
  • medo
  • prometheus
  • react
  • row-level-security
  • slack-webhooks
  • supabase-auth
  • supabase-database
  • supabase-edge-functions
  • tailwind-css
  • telegram-bot-api
  • typescript
Share this project:

Updates

Submission history