Get Started with Lakehousecat

Lakehousecat is an AI-powered analytics platform that connects your data to large language models — enabling natural language queries, automated chart generation, and semantic data modeling in your own infrastructure.

Choose Your Path

Evaluate Locally — TUI Installer (Recommended)

The fastest way to try Lakehousecat is the TUI Installer (lhc-installer). It sets up a complete local instance automatically — no Kubernetes experience required.

macOS / Linux

curl -fsSL https://install.lakehousecat.com | bash

Windows (PowerShell 5.1+)

iex (iwr https://install.lakehousecat.com/install.ps1).Content

The installer guides you through system checks, portal authentication, and automated deployment. Deploying the full service stack takes about 10 minutes, depending on your internet bandwidth.

Minimum system requirements:

• macOS, Linux, or Windows (10/11, amd64 or arm64)
• 32 GB RAM, 10 CPU cores, 50 GB free disk (available to Docker)
• Docker Desktop (Windows/macOS) or Docker Engine (Linux), and Minikube

→ TUI Installer Guide

---

Deploy to Kubernetes

For production deployments or integration into an existing cluster, use the Kubernetes Operator via Helm. This path requires a Kubernetes cluster and familiarity with Helm and kubectl.

helm repo add lakehousecat https://charts.lakehousecat.com
helm repo update

# The operator only gets rights in the instance namespaces listed in
# targetNamespaces and does not create them: create the namespace first.
kubectl create namespace lhc-instance

helm install lhc-operator lakehousecat/lakehousecat-operator \
  --namespace lhc-operator \
  --create-namespace \
  --set "targetNamespaces={lhc-instance}" \
  --wait

Order of steps: register at portal.lakehousecat.com (free) first to obtain your Handshake Key, then deploy your Lakehousecat instance. The operator itself (the Helm install above) can be installed at any time, but the Handshake Key is required before the operator can provision an instance.

→ Kubernetes Deployment Guide

---

First Login: Your Role as Administrator

When Lakehousecat is first deployed, you log in as the Administrator. The system creates the initial admin account automatically during deployment — credentials are stored in a Kubernetes secret. Keep that secret's API key accessible outside the cluster (a password manager, not another resource only the cluster can reach) — it doubles as your recovery path if you ever lose access to MFA on this account; see Locked out of MFA?.

The Administrator is the only role that exists at this point. As Administrator, your responsibilities before handing the system over to other users are:

Task	Why
Configure at least one Provider Model	Required for all AI operations — sessions, semantic extraction, and chart generation all need an LLM
Connect and load Data Sources	Data must be in ClickHouse before it can be analyzed
Run Semantic Extraction	Builds the intelligence layer on top of the raw data
Create Custom Models	Defines what data users can query and which LLM they use
Invite users and assign roles	Builders and Users cannot self-register by default
Share Custom Models	Users only see models explicitly shared with them

Until a Custom Model exists and is shared, no regular user can do anything meaningful in the system. The Administrator sets up the platform before handing it over.

For a complete step-by-step walkthrough of this setup process, see the Quickstart Guide.

For a detailed explanation of all three roles (Administrator, Builder, User), see Roles.

---

First Steps After Installation

Once Lakehousecat is running, follow these steps to start answering questions with your own data:

1. Access the UI

Retrieve your admin credentials from the Kubernetes secret created by the Operator:

kubectl get secret lhc-admin-secret -n lhc-instance \
  -o jsonpath='{.data.LHC_ADMIN_EMAIL}' | base64 -d && echo

kubectl get secret lhc-admin-secret -n lhc-instance \
  -o jsonpath='{.data.LHC_ADMIN_PASSWORD}' | base64 -d && echo

Open http://localhost:42021 and log in.

2. Connect a Data Source

Navigate to Data → Data Sources → + and connect your first data source.

Supported types: PostgreSQL, MySQL, Microsoft SQL Server, Delta Lake, Apache Hudi, Apache Iceberg, Amazon S3, File Upload

→ Connect a Data Source

3. Build the Semantic Layer and Create a Custom Model

After connecting a data source, trigger semantic extraction to analyze the schema and generate descriptions for tables and columns. This is what enables the AI to understand your data structure.

Navigate to the data source editor → Operations tab → Create Semantic Layer.

Then create a Custom Model: navigate to Models → Custom Models → +, select a provider model (OpenAI, Anthropic, Google, Azure, AWS, or xAI), and link the data source. The semantic layer of the linked data sources becomes the knowledge base for the model.

→ Create a Model Guide

4. Start a Session

Open Sessions, select your Custom Model, and ask a question in natural language. Lakehousecat generates SQL, executes it against your data source, and returns an answer with a visualization.

→ Sessions Guide

---

Subscription Tiers

Tier	Users	Data Sources	Support
FREE	1	10	Portal
STANDARD	Up to 10	50	Portal
PREMIUM	Up to 100	Unlimited	Priority
ENTERPRISE	Individual	Unlimited	Dedicated

All tiers run on the same platform. Upgrade through the portal — no redeployment required.

---

Learn More

• Guides — Step-by-step workflows for every feature

Support

• portal.lakehousecat.com — Feature requests, support tickets, and account management
• Priority Support — Included with STANDARD, PREMIUM, and ENTERPRISE tiers