ClickHouse

Your logs are talking. Start listening.

Build a real-time log analysis dashboard from your ClickHouse data. Severity heatmaps, error rate trends, and natural-language log queries—built by AI in one conversation.

tail -f /var/log/services/*.log
2026-03-28 09:14:02.118[INFO]Request served 200 /api/v1/users 12ms
2026-03-28 09:14:02.445[INFO]Health check OK pid=3201
2026-03-28 09:14:03.012[WARN]Slow query detected duration=1.2s table=events
2026-03-28 09:14:03.334[ERROR]Connection refused host=db-replica-03:9000
2026-03-28 09:14:04.118[INFO]Request served 200 /api/v1/orders 8ms
2026-03-28 09:14:05.221[ERROR]OOM killed pid=4821 rss=2.1GB
2026-03-28 09:14:05.890[INFO]Request served 200 /api/v1/health 2ms
2026-03-28 09:14:06.102[DEBUG]Cache hit key=user:8821 ttl=340s
>

Logs are the most valuable data you never look at.

Your infrastructure generates millions of log lines every hour. They contain the answers to every outage, every slow endpoint, every silent failure. But extracting those answers? That is where things fall apart. You are drowning in data and starving for insight.

Grep is not observability

You are SSHing into boxes and grepping gigabyte log files. By the time you find the needle, the haystack has doubled. There is no pattern recognition, no trend analysis, no cross-service correlation.

Alert fatigue kills response time

Your monitoring fires 200 alerts a day. Most are noise. By the time you triage the real ones, the incident is 20 minutes old and your users already noticed.

ELK and Datadog costs spiral at scale

You are paying thousands per month to index logs you rarely query. The dashboards take a week to set up, another week to tune, and break every time someone changes a log format.

Ask your logs anything. In English.

camelAI connects to your ClickHouse instance, learns your log table schemas, and builds a full-featured analysis dashboard. Query billions of log lines in natural language. The agent writes the ClickHouse SQL, formats the results, and deploys a live dashboard you can share with your team. No Grafana JSON, no Kibana config files, no vendor lock-in.

U
Show me all ERROR logs from the payments service in the last hour, grouped by error message
AI

Found 47 errors across 6 unique messages. Here is the breakdown:

Error MessageCount
Connection refused18
Timeout exceeded12
OOM killed8
Auth failed5
Disk full3
Schema mismatch1
Natural language queriesAuto-generated ClickHouse SQLLive deployed dashboard

What you will build.

log-dashboard--acme.camelai.app

Total Events

0.0M

+8% vs yesterday

Errors (24h)

0

+23% vs yesterday

p99 Latency

0ms

-15ms vs yesterday

Uptime

0.0%

steady

Error Rate (24h)

00:0006:0012:0018:0023:59

Severity Heatmap

000408121620
MonTueWedThuFriSatSun
TimestampSeverityServiceMessage
09:14:05.332ERRORpaymentsOOM killed pid=4821 rss=2.1GB
09:14:05.118INFOgatewayRequest served 200 /api/v1/users 12ms
09:14:04.891WARNauthSlow query detected duration=1.8s
09:14:04.445INFOpaymentsRequest served 200 /api/v1/orders 8ms
09:14:03.998ERRORdbConnection refused host=replica-03:9000
09:14:03.772DEBUGcacheCache hit key=user:8821 ttl=340s

Three steps. One conversation.

1

Connect ClickHouse

Add your ClickHouse connection details in camelAI's integrations panel -- host, port, database, and credentials. The agent discovers your log tables and column schemas automatically. Supports ClickHouse Cloud, self-hosted, and Altinity deployments.

logs_production

ClickHouse Cloud

Connected
2

Describe what you need

Tell the agent exactly what you want in plain English.

“Build me a log analysis dashboard with error rate charts by service, a severity heatmap by hour, and a searchable log viewer with filtering by service, severity, and time range.”
3

Deploy and monitor

The agent builds your dashboard with React, Recharts, and Tailwind -- then deploys it to a live URL. Set up a cron job to check error rates every 5 minutes and post to Slack when they spike above your threshold.

log-dashboard--acme.camelai.appLive

Queries that used to take 20 minutes of SQL. Now they take one sentence.

Show error spikes in the last 6 hours

Result: Sparkline

Which services have the highest error rate this week?

Result: Ranked bar chart

payments
auth
gateway

Find all timeout errors from the auth service after the last deploy

Result: Filtered table

14:02:11timeout
14:02:38timeout
14:03:05timeout

Compare today's log volume to the same day last week

Result: Overlay comparison

What was the p99 query latency yesterday vs today?

Result: Comparison stat

Yesterday

357ms

Today

342ms

-4.2%

Show me the full stack trace for request ID abc-123-def

Result: Trace waterfall

gateway
auth
db

Built for the people who get paged at 3 AM.

DevOps Engineers

You run the infrastructure. You need to see what is happening across 50 services without opening 50 tabs. Build a single pane of glass over your ClickHouse logs.

SREs

You own uptime. You need to see error patterns before they become incidents. Get a dashboard that highlights anomalies automatically and helps you cut MTTR.

Platform Engineers

You build the internal tools your team uses every day. Give them a log dashboard that actually helps -- searchable, filterable, and deployed to a URL they can bookmark.

Frequently asked questions

Stop grepping. Start seeing.

Connect ClickHouse, describe the dashboard you need, and deploy it in minutes.