Product Update - 🤖 Introducing KAI Assistant

An AI assistant built directly into Kleene is now live.

KAI Assistant lets you interact with your data, transforms, and documentation using plain English — no SQL from scratch, no hunting through docs. It's context-aware, role-aware, and built with enterprise data privacy at its core.

Let's dive in 👇

⚙️ Enabling KAI Assistant

To enable KAI Assistant, navigate to App Settings → AI tab. From there you can switch it on and configure what it can access.

💬 What KAI Assistant supports

• Guidance — Questions about Kleene features, best practices, and understanding data pipelines. KAI Assistant searches the documentation for relevant information.
• Transform operations — Transform and group search, searching logs for a specific transform, previewing table data, and generating code.
• Troubleshooting — Explaining error messages and suggesting fixes to transforms.

🚪 Accessing KAI Assistant

Two ways to open KAI Assistant:

a) Click the icon in the bottom right of any app page b) Click the ... menu on any transform → Ask AI — this gives KAI Assistant the context of that individual transform for more specific responses.

✏️ Prompt best practices

When KAI Assistant opens without an active conversation, suggested prompts appear to help guide you toward useful answers or specific workflows.

Clarity is key. Use straightforward language and direct commands — the more context you provide, the more tailored the response. For example:

• "How do I set up an Asana source?"
• "How do I optimize a transform schedule?"

🔐 Data privacy

KAI Assistant is built opt-in from the ground up. You control what it can access via App Settings → AI tab.

• For general Q&A, KAI Assistant only uses docs and your prompt — no warehouse data involved.
• For schema and preview workflows, KAI Assistant generates synthetic sample data by default (via AWS Bedrock), stripping PII, passwords, and card numbers before anything reaches the LLM.
• If you enable raw data previews, those previews are not used for model training and are not logged by third-party services.
• KAI Assistant runs on Google Gemini via Vertex AI — your data is never used to train the Gemini model.

🎯 Why this matters

Less time writing SQL. Less time searching docs. Faster onboarding for new team members. KAI Assistant lowers the barrier to working with your data — for engineers and analysts alike.

Read the full blog post here or explore the docs at docs.kleene.ai.

Questions? Reach out to your CSM anytime.