ClickHouse

Surface ClickHouse analytics insights through natural customer conversations

Billions of rows queried in milliseconds, now accessible through a simple chat. Your AI agent translates customer questions into ClickHouse SQL, retrieves real-time analytics, and delivers precise answers. No dashboards needed, no data team bottleneck.

Chosen by 800+ global brands across industries

Real-time OLAP queries at conversation speed

Your AI agent becomes a conversational analytics layer on top of ClickHouse, turning plain-language questions into blazing-fast columnar queries and human-readable responses.

ClickHouse

Use Cases

Analytics conversations, not dashboards

See how businesses let customers and internal teams query billions of data points through natural language conversations powered by ClickHouse.

Self-Service Usage Reporting for SaaS Customers

A SaaS customer asks 'How many API calls did we make last month?' Your AI Agent translates that into a ClickHouse SQL query against your usage tracking tables, aggregates the data across all endpoints and dates, and responds with a formatted breakdown. The customer gets an instant answer without filing a support ticket, and your data team stays focused on building, not reporting.

Real-Time Product Analytics on Demand

A product manager messages 'What was our conversion rate this week compared to last week?' The AI Agent queries your ClickHouse events table, calculates both weekly conversion rates, computes the percentage change, and presents the comparison clearly. Decisions happen in seconds instead of waiting for the next analytics review meeting.

Instant Anomaly Investigation for Operations

An ops engineer notices a spike in error rates and asks 'Which endpoints had the most 500 errors in the last hour?' Your AI Agent runs a grouped aggregation query on your ClickHouse logs table, ranks endpoints by error count, and surfaces the top offenders with timestamps. Root cause analysis starts immediately, cutting mean-time-to-resolution significantly.

Try
ClickHouse

ClickHouse

FAQs

Frequently Asked Questions

What types of SQL queries can the AI agent run on ClickHouse?

The agent can execute any read query supported by ClickHouse, including SELECT statements with aggregations, joins, window functions, and subqueries. It uses the Execute Query endpoint with configurable output formats like JSONEachRow, CSV, or TSV. Write queries can be restricted by setting the readonly=1 parameter in query settings.

How does the agent know which tables and columns to query?

Before constructing queries, the agent calls the Get Database Schema and Get Table Schema endpoints to discover available tables, column names, data types, and primary keys. It maps your question to the correct schema automatically. You can also train the agent on your specific data dictionary for improved accuracy.

Does Tars store any data from my ClickHouse database?

No. Every query is executed in real-time against your ClickHouse instance, and results are used only to formulate the current conversation response. Tars does not cache, replicate, or persist any of your ClickHouse data. Query results exist only for the duration of that conversation turn.

Can I limit query execution time to prevent expensive long-running queries?

Yes. The Execute Query endpoint supports a max_execution_time parameter measured in seconds. Set it to 30 or 60 seconds to prevent runaway queries. You can also pass additional ClickHouse settings like max_rows_to_read to cap the volume of data scanned per request.

Does this work with ClickHouse Cloud and self-hosted instances?

Both. Provide your ClickHouse Cloud URL (typically https://your-cluster.clickhouse.cloud:8443) or your self-hosted HTTP interface address. Authentication uses the same username and password credentials regardless of deployment model.

How is this different from connecting a BI tool like Metabase or Grafana to ClickHouse?

BI tools require users to build dashboards and write queries. Tars lets anyone ask questions in plain language and get answers from ClickHouse without knowing SQL, without navigating a dashboard, and without waiting for a data analyst. It is conversational analytics, not static reporting.

Can the agent query across multiple ClickHouse databases in one conversation?

Yes. The agent can list all databases, switch between them using the database parameter on each query, and correlate results across multiple databases within a single conversation thread. It discovers available databases dynamically using the List Databases endpoint.

What permissions should I give the ClickHouse user account that Tars connects with?

For read-only analytics, create a ClickHouse user with SELECT privileges on the relevant databases and tables. Avoid granting INSERT, ALTER, or DROP permissions. You can use ClickHouse's role-based access control to restrict which databases the Tars user can access, limiting exposure to sensitive tables.

How to add Tools to your AI Agent

Supercharge your AI Agent with Tool Integrations

Don't limit your AI Agent to basic conversations. Watch how to configure and add powerful tools making your agent smarter and more functional.

Privacy & Security

We’ll never let you lose sleep over privacy and security concerns

At Tars, we take privacy and security very seriously. We are compliant with GDPR, ISO, SOC 2, and HIPAA.

GDPR
ISO
SOC 2
HIPAA

Still scrolling? We both know you're interested.

Let's chat about AI Agents the old-fashioned way. Get a demo tailored to your requirements.

Schedule a Demo