For operations teams

Investigate operational issues with saved analysis

When something breaks, trace it back to the data. Use saved queries, notebooks, and dashboards to investigate incidents and prevent repeat issues.

Teams: Ops & SRE
Speed: Query in seconds
History: Saved investigations
Agent: Kole assists

When an operational issue hits, the first question is always "what changed?" The second is "has this happened before?" SyneHQ gives operations teams a single place to trace issues back to the data, save the investigation, and hand off findings without rebuilding the query from scratch every time an alert fires.

Instead of jumping between monitoring dashboards, log platforms, and chat threads, your team works from one environment: query the data directly, annotate the findings, and keep the investigation where the next on-call engineer can find it. The goal is not to replace your alerting stack but to give the response a home that outlasts the incident channel.

Investigate from alert to root cause

The investigation workflow follows a predictable path. An alert fires, someone opens a query, and the team traces backward through the data until the root cause is visible. SyneHQ structures this path so each step is recorded and reusable.

Start with a saved query or open a new one in Data Explorer. Pull the relevant time window, filter to the affected service or entity, and let the results guide the next question. When the investigation requires more than a single query, move into Quantum Lab and build an ordered notebook: SQL blocks for the raw data, Python blocks for transformations, and Markdown blocks to capture what you are seeing and why it matters.

Kole, the AI agent, can assist at each stage. Point it at the relevant connection, describe the issue, and let it propose queries against your schema. It reads the available tables, inspects column types, and suggests a starting point. You review and approve each proposed action before it runs, so the investigation stays under your control.

Once the root cause is identified, save the notebook. The next time a similar alert fires, your team opens the saved investigation instead of starting from zero.

Operations / investigation workflow

From alert to root cause in four steps.

Illustrative workflow

Each step is recorded in a Quantum Lab notebook so the next on-call engineer can retrace the investigation.

Alert triggered

Processing latency exceeds threshold at 14:32 UTC.

Kole queries logs

Agent proposes a diagnostic query against processing_logs.

Root cause identified

Payment verification step slowed after a 14:30 deployment.

Saved for next time

Notebook with queries and finding stored for the team.

Kole-proposed diagnostic query

SELECT event_type, COUNT(*), AVG(duration_ms)
FROM processing_logs
WHERE created_at > NOW() - INTERVAL '4 hours'
GROUP BY event_type
ORDER BY COUNT(*) DESC

Sample query - review and approve before execution

Operational metrics at a glance

Not every operational question starts with an incident. Teams also need a standing view of how systems are performing: processing rates, error rates, latency percentiles, and throughput over time. SyneHQ Dashboards let you build these views from the same queries your team uses during investigations.

A dashboard can combine chart blocks for time series data with Markdown blocks that explain thresholds, ownership, and escalation paths. Pin the queries that matter to a shared dashboard, set a refresh interval, and let the on-call engineer open one page instead of running five queries manually.

The value compounds over time. Each query you write during an investigation becomes a candidate for a dashboard panel. Error rate spikes that triggered a manual investigation last quarter become a chart that the team watches in real time this quarter.

Operations / dashboard

Standing metrics for the on-call engineer.

Processing Rate

0/min

Error Rate

P95 Latency

0ms

Hourly processing volume (sample)

Two-hour windows, events per minute - sample values only

Patterns operations teams recognize

Operations work follows patterns. The specific systems vary, but the shape of the investigation is consistent. Here are concrete examples of how teams use SyneHQ to handle recurring operational scenarios.

Order processing delays

A spike in order processing time triggers an alert. The on-call engineer opens a saved query that breaks processing duration by event type and time window. The query shows that a downstream payment verification step slowed from 200ms to 1,400ms starting at 14:32 UTC. The engineer traces the timestamp to a deployment, rolls it back, and updates the saved notebook with the finding. Next time the alert fires, the team checks the same notebook first.

Inventory sync failures

A nightly inventory sync job fails silently, and the team discovers the gap the next morning when fulfillment reports a mismatch. Using Data Explorer, the engineer queries the sync log table for the last 72 hours, filters by status, and finds that a schema change in the upstream system broke the field mapping. The investigation is saved as a Quantum Lab notebook with the diagnostic queries, the root cause note, and a recommended check to add to the sync job.

Integration health checks

The team monitors a set of third-party API integrations that feed data into the warehouse. A dashboard tracks response times, error codes, and payload sizes by integration partner. When one partner's error rate crosses a threshold, the on-call engineer opens the dashboard, clicks into the underlying query, and pivots to a notebook investigation. The dashboard catches the signal; the notebook captures the diagnosis.

Infrastructure cost anomalies

A weekly query compares compute and storage costs against the previous period. When costs jump unexpectedly, the team traces the increase to a specific service or query pattern. The saved investigation includes the cost query, the attribution breakdown, and the action taken. Over time, the team builds a library of cost investigations that new members can reference during onboarding.

Keep the investigation with the team

The most expensive part of an operational issue is not the fix. It is the time spent rediscovering what happened, why, and what was done about it. When investigations live in chat threads or personal notebooks, that context evaporates.

SyneHQ keeps investigations in a shared, searchable environment. Saved queries in Data Explorer, ordered notebooks in Quantum Lab, and standing dashboards all belong to the team. When someone rotates off on-call, the next person inherits the same tools, the same saved investigations, and the same dashboards.

Kole accelerates the routine parts: proposing queries, inspecting schemas, and surfacing relevant saved work. The human operator still decides what to investigate, what the data means, and what action to take. The agent handles the lookup; the engineer handles the judgment.

Start by saving your next investigation. Open Quantum Lab to build an ordered notebook for an active issue, or set up a Dashboard to track the metrics your team checks every day.