





Job failed? Open Kai, type "Fix it." With follow mode enabled, Kai navigates you through the pages it's working on so you see changes in real-time. Disable it to let Kai work in the background.
We wanted to build something different - a real data engineering agent capable of managing the entire lifecycle of data processing within Keboola.
Guided Workflows
Job failed? Open Kai, type "Fix it." With follow mode enabled, Kai navigates you through the pages it's working on in real-time.
Full Transparency
Kai asks before performing any modifying action and displays a diff with proposed changes. Every action is logged in a complete audit trail.
Safe Development
Kai can work in development branches, ensuring it never affects production. Review all changes before merging.
Kai lives inside your Keboola project. It sees your schemas, reads your logs, and understands your pipelines — so it fixes real problems, not hypothetical ones.
Jordan Burger
AI Lead at Keboola / Your guide to Kai AI Assistant
Your data is never used to train AI models . Full audit trails, regional processing, and automatic PII redaction.
Your data is never used to train AI models. Zero retention on processed data.
Every Kai action is logged with complete traceability for org admins.
Data stays in your region. GDPR, HIPAA, SOC 2 Type II compliant.
Automatic detection and redaction of personally identifiable information.
Granular permissions control who can use Kai and what it can do.
Credentials and tokens are never exposed to Kai. Safe secrets handling.
When something breaks, Kai helps you fix it fast.It can analyze job failures, resolve configuration errors, investigate data quality issues, and debug API authentication problems.
Kai can build complete data apps from scratch — dashboards, reports, and interactive tools — directly within Keboola, without needing a separate dev environment.
Kai writes custom connectors and configures integrations on demand. Describe what you need to connect — Kai handles the setup, authentication, and mapping.
Tell Kai what transformation you need — it writes, tests, and deploys SQL or Python code directly into your pipeline. No context switching needed.
New to the project? Kai reads your entire Keboola setup and explains what every component does, how data flows, and where to start — in plain language.
Kai auto-generates documentation for your tables, transformations, and pipelines — and keeps metadata up to date as your project evolves.
No installation, no API keys, no OAuth setup. Kai is already embedded and authenticated in your project.
Find Kai in your project's top navigation bar. It's already context-aware of your current location — whether you're on Flows, Storage, or a specific component.
Describe what you need. Kai reads your schemas, logs, configurations, and pipelines to understand the full picture before responding.
Every modifying action shows a code diff and requires your explicit consent. Test in development branches for safety, then merge to production when satisfied.
If you paste an error log into ChatGPT, it guesses based on syntax. Kai reads the actual state of your project and fixes based on what's actually there.
Can I use both? Yes! They are not mutually exclusive. Many users rely on Kai for quick troubleshooting and the MCP Server with Cursor or Claude Desktop for heavy-duty local development.
Whether you're exploring data, building integrations, debugging pipelines, or generating documentation — Kai is ready to help.