ClickHouse + ASI Biont: How an AI Agent Automates Analytics Without SQL Queries and Dashboards

Introduction: Real-Time Analytics Without Pain

ClickHouse is a columnar analytical DBMS created by Yandex in 2016. It is designed for real-time query processing on large volumes of data—from gigabytes to petabytes. According to official documentation (ClickHouse Docs, 2024), the database is used by companies such as Cloudflare, Uber, eBay, and Bloomberg for log analysis, metrics, event tracking, and advertising analytics. However, as data volumes grow, a problem arises: analysts and developers have to write complex SQL queries to extract useful insights, and business users are often cut off from data until they receive a ready-made report from the team.

This is where the ASI Biont AI agent comes to the rescue. Instead of spending hours writing queries and setting up dashboards, you can simply communicate with the agent in natural language—and get answers, visualizations, and automatic updates. In this article, I will explain how to set up the ClickHouse integration with ASI Biont, what tasks it solves, and why it changes the approach to working with data.

How ASI Biont Connects to ClickHouse: No Panels, Only Chat

Unlike traditional automation platforms (Zapier, n8n), where you need to find a ready-made integration module, configure triggers and fields, ASI Biont uses a fundamentally different approach. The user simply provides an API key from the service in a dialogue with the AI agent. After that, the AI itself writes the integration code for your service's API—in this case, ClickHouse.

ClickHouse supports several protocols for third-party access: HTTP REST API (port 8123), Native TCP (port 9000), and a PostgreSQL-compatible interface (since version 24.6). ASI Biont uses the HTTP REST API—the simplest and most secure connection method. The agent automatically recognizes which tables, schemas, and fields are available and offers usage scenarios.

What You Need to Do:

1. Open a chat with ASI Biont at asibiont.com.
2. Write: "Connect my ClickHouse database. Here is the API key: [your key]." If you don't have a dedicated API key, you can provide a login/password or URL with parameters—the AI will figure it out.
3. The agent will test the connection, display a list of tables, and ask what data you want to analyze.
4. After that, you simply ask questions in natural language, and the AI turns them into SQL queries, executes them, and returns the answer.

What Tasks This Integration Automates

The integration with ClickHouse via ASI Biont covers three key areas: data collection, real-time analysis, and visualization.

1. Automatic Data Collection and Aggregation

ClickHouse is often used to store event logs—for example, user clicks, server requests, transactions. Instead of manually writing queries for daily reports, you can set up the AI agent to run queries regularly and send results.

Example: You tell the agent: "Every morning at 9:00 AM, send me the number of unique users from yesterday, grouped by traffic source." ASI Biont will create the corresponding SQL query (using GROUP BY and countDistinct), execute it, and send the result to the chat or email.

2. Answering Business Questions in Natural Language

This is the most powerful feature. Business users don't need to know SQL—they just ask questions like they would to a colleague analyst.

Examples of questions the agent understands:
- "Show the top 10 pages by load time in the last hour"
- "Compare purchase conversion between mobile and desktop traffic for July 2026"
- "Which products are most often bought together?" (requires a JOIN with order tables)

ASI Biont translates the request into correct SQL considering the data schema, executes it in ClickHouse, and returns the answer as a table, chart, or text description.

3. Visualization Without Dashboards

You don't need to install Grafana, Metabase, or another visualization tool. ASI Biont can build charts directly in the chat: line charts, histograms, pie charts, and heat maps. The agent automatically selects the visualization type based on the data.

Scenario: You say: "Build a line chart of the number of orders by day for the last 30 days." The AI executes a query to ClickHouse using the toDate function for grouping and returns an interactive chart that can be opened in a browser or downloaded.

Practical Use Cases

Scenario 1: DevOps Log Analytics

Imagine you are an engineer at a company that uses ClickHouse to store web server logs (e.g., Nginx). Millions of records are generated daily. Previously, to find errors, you wrote queries manually:

SELECT status, count(*) as cnt
FROM logs
WHERE timestamp >= now() - INTERVAL 1 HOUR
  AND status >= 500
GROUP BY status
ORDER BY cnt DESC;

With ASI Biont, you simply write: "How many 5xx errors in the last hour? Show the distribution by code." The AI will generate the query, execute it, and send the answer. If needed, you can set up automatic alerts when a threshold is exceeded.

Scenario 2: Real-Time Marketing Analytics

A marketer wants to track the effectiveness of an advertising campaign. ClickHouse stores data on clicks, impressions, and conversions from ad networks. Instead of waiting for a report from an analyst, the marketer asks the agent: "Which creatives have the best CTR by gender today?" ASI Biont executes a query with grouping by creative and gender, using anyHeavy or uniqExact for accurate counting, and returns a table with results.

Scenario 3: Financial Monitoring

The finance department uses ClickHouse to store transactions. They need to quickly find out which expense categories have grown over the week. Question: "Show the top 5 categories with the largest increase in transaction amount compared to last week." The AI generates a query with lagInFrame or self-join for period comparison, executes it, and displays the result. Time saved: from 30 minutes to several hours per day.

Why It's Beneficial: Time and Resource Savings

According to a McKinsey survey (2024), analysts spend up to 40% of their working time on data preparation and writing queries. ASI Biont automates this routine, allowing you to focus on interpreting results and making decisions.

How to Get Started: Step-by-Step Instructions

1. Get access to ClickHouse. Make sure your ClickHouse server is accessible via HTTP. If you are using a cloud service (e.g., ClickHouse Cloud), copy the URL and password from the settings.
2. Go to the chat with ASI Biont at asibiont.com.
3. Provide connection details. Write: "Connect ClickHouse. URL: http://your-server:8123, user: default, password: [password]." The agent will check the connection and show a list of tables.
4. Start asking questions. For example: "Show the structure of the events table," "How many records are in the sales table?"
5. Set up automation. If needed, ask the agent to run a specific query every day at 10:00 AM and send the result.

Conclusion

ClickHouse is a powerful tool for analytics, but its potential often remains untapped due to the complexity of SQL and the need to involve analysts for every query. Integration with the ASI Biont AI agent removes this limitation: any company employee can ask a question in natural language and instantly get an answer with visualization. Setup takes minutes—just provide the API key in the chat, and the AI will write the integration code.

Try this capability right now. Go to asibiont.com, open the chat, and connect your ClickHouse database. See for yourself that analytics can be fast and accessible to everyone.