Introducing KAI Assistant: Natural Language SQL and Pipeline Intelligence Inside Kleene.ai
February 17, 2026
— min read
At the end of Q1, we’re launching KAI Assistant – a native AI assistant built directly into the Kleene.ai data platform.
Phase 1 focuses on what teams use every day: SQL transforms, pipeline navigation, logs, and documentation search. Future phases will go further, including pipeline creation and analytics model context.
If you have ever searched through transforms, checked logs manually, or written repetitive SQL from scratch, KAI Assistant changes that workflow immediately.
"Kleene.ai has long established that predictive models powered by Ai only deliver value when the underlying data is robust and fully integrated. We will shortly be integrating Kai with our suite of machine learned models for pricing, media optimization, forecasting, segmentation, demand, inventory management and creative diagnostics. Our model orchestration layer will monitor all active models in real time and measure where incremental gains or cost savings are being generated. Kai will enable users to query these models directly without the need for further analytics, interpretation or visualization”
Ian Liddicoat, Kleene.ai CTO
What Is KAI Assistant?
KAI Assistant is a native natural language AI built directly into Kleene.ai, powered by Google Vertex AI and the latest Gemini models.
It allows you to:
Generate and optimize SQL transforms
Search transforms and groups
Fetch table schemas and preview table data
Search logs for specific transforms
Interactively query Kleene documentation
Generate synthetic data previews for testing
All from within the platform.
Instead of switching between SQL console, logs, docs, and Slack threads, you ask KAI.
What’s Included in Phase 1
Phase 1 focuses on transform intelligence and documentation awareness.
Transform Intelligence
You can:
Generate SQL from natural language prompts
Optimize existing transforms
Search transform groups
Fetch transforms within a group
Retrieve the latest version of a transform
Search logs for a specific transform
This makes debugging and iteration significantly faster.
Table Awareness
KAI Assistant can:
Fetch table schemas
Preview table data
Generate synthetic data for sample previews
Assist with parsing and transformation suggestions
When sample data is used, it is converted to synthetic data before any LLM processing.
Documentation Intelligence
KAI Assistant supports RAG-style search across docs.kleene.ai.
You can ask:
“How does pipeline scheduling work?”
“What’s the recommended pattern for incremental transforms?”
“How do I enable webforms ingestion?”
And get contextual answers grounded in our documentation.
KAI Assistant refining SQL to optimize a transform
What’s Coming Next
Phase 2
Ability to create full ELT pipelines Extract → Transform → Data Product
Deeper vertical integration
Context about statistics and analytics outputs
Phase 3
Analytics model context
Composite values from AI models
Natural language interaction with model outputs
In the near future, KAI Assistant will allow users to query KAI Analytics model results through natural language prompts. That means asking questions about forecasting, segmentation, or optimization outputs directly through the KAI Assistant.
Who It’s For
KAI Assistant is designed for both technical and non-technical users.
For data teams
Faster SQL generation
Reduced debugging time
Easier pipeline navigation
Log and transform search without manual digging
For analytics managers and business users
Plain language interaction
Easier access to metadata
Documentation answers without searching manually
How It Works (High-Level Architecture)
KAI Assistant operates within Kleene.ai and interacts securely with AI services. Gemini is accessed through Google Vertex AI, Google Cloud’s managed enterprise AI platform, ensuring secure and governed model access.
There are two pathways:
Kleene.ai → Gemini via Google Vertex AI (standard interactions)
Kleene.ai → Gemini via Google Vertex AI (when synthetic sample data preview is enabled)
We do not store or use customer data to train any models. All processing follows Kleene.ai’s security standards and governance policies.
What Changes vs Today?
Before, everyday reporting tasks often meant jumping between tools and doing things manually. Writing SQL from scratch, searching through transform groups to find the right version, digging into logs to debug issues, and opening documentation in another tab were all part of the workflow. Each step was small, but together they slowed teams down and pulled focus away from delivering insight.
With KAI Assistant built natively into the Kleene.ai data platform, that friction disappears. Teams can generate SQL using natural language, instantly locate transforms, debug issues with context-aware suggestions, and get documentation answers without ever leaving the interface. The result is faster reporting cycles, fewer bottlenecks, and a workflow that feels fully integrated with your data platform rather than stitched together across multiple tabs.
The Big Picture
KAI Assistant is not just a SQL helper, it is the foundation for:
Natural language querying of analytics models
AI-guided pipeline creation
Context-aware optimization suggestions
Integrated decision intelligence inside Kleene.ai
As future phases are released, KAI Assistant will move from just assisting with transforms to understanding full analytics models and business outputs.
KAI Assistant launches in Q1, with opt-in access rolling out shortly after release.
If you are already using Kleene.ai, get ready to write less SQL and ship transforms faster. If you are evaluating modern data platforms, this is what an AI-native workflow looks like.
If you want to see KAI Assistant in action, book a demo with our experts.
%201.svg)
