ResolveOps AI

Architecture

How ResolveOps AI ingests operational signals and stores them as queryable incident events.

Processing pipeline
From raw operational events to resolved incidents.

1. Ingest

Normalize events from PagerDuty, Datadog, CloudWatch, Slack, Jira, and Zendesk into a unified schema.

2. Correlate

Store and query incident signals in Amazon DynamoDB using incidentId and eventTimestamp keys.

3. Analyze

AI ranks root-cause hypotheses, scores recurrence risk, and cites evidence from the timeline.

4. Resolve

Generate postmortems and prevention action items, tracked to closure across owners and services.

DynamoDB event model
Core incident and event records stored in Amazon DynamoDB table resolveops-ai-events. Incidents connect services, timelines, responders, root causes, postmortems, and action items.
TablePurposeColumns
organizationsTenant accounts and plan metadata7
usersResponders, owners, and commanders9
servicesTracked services and health state11
incidentsCore incident records14
incident_servicesMany-to-many incident ↔ service join4
incident_timeline_eventsDetection → resolution event log8
incident_respondersResponder roles per incident6
postmortemsStructured postmortem documents12
root_causesRoot cause + confidence per incident7
action_itemsFollow-up actions and prevention11
external_eventsIngested events from data sources9
ai_recommendationsAI summaries and recommendations8
audit_logsImmutable change history7
Connected data sources
Signals feeding the external_events table.
PagerDutyAlerting
connected
SlackCollaboration
connected
JiraTicketing
connected
AWS CloudWatchMonitoring
connected
DatadogMonitoring
simulated
ZendeskSupport
simulated
Tech stack
The foundation ResolveOps AI is built on.
Next.js App RouterAmazon DynamoDBServer ComponentsAI root-cause analysisEvent correlation engineRole-based incident command