Skip to content

Forewarden presents Lakehouse Studio

Lakehouse Studio gives your team SQL, notebooks, dbt, jobs, dashboards and an AI assistant in one workspace that runs on servers you control.

Free for 30 days on your own server. One command to install. No card needed.

22
AI model providers
OpenAI, Anthropic, Ollama, vLLM and 18 more behind one gateway
5
SQL engines
DuckDB, PostgreSQL, Spark, chDB and Unity Catalog
6
SQL functions for documents
Parse, extract, classify and decide from a query
4
AI governance templates
Open, Internal, Confidential and Regulated

Work with data

Write it, chart it, schedule it.

SQL, Python and dbt in one workspace, with dashboards built from the same SQL.

  • SQL editor and notebooks

    Write SQL with autocomplete, parameters, charts and saved versions. Notebooks mix SQL, Python and markdown, and each person gets a private Python process.

  • Dashboards and alerts

    Build dashboards from SQL datasets, publish them and refresh them on a schedule. Alerts run a query on a schedule and notify people when a condition holds.

  • dbt Core projects

    Run dbt Core from the app with live logs, run history and generated documentation. Models show up in lineage and can run as a job task.

  • Visual Data Prep

    Build data pipelines on a canvas, preview every step and save the result as a table, a view or a dbt model.

Catalog and lineage

Know what you have and where it came from.

Browse every table, follow it to its sources and to everything that reads it, and share it when you decide to.

  • Catalog and Delta tables

    Browse catalogs, schemas, tables and volumes. Delta tables keep their versions, so you can read an earlier one with VERSION AS OF or a timestamp.

  • Unity Catalog and shared metastores

    Run the open source Unity Catalog server and attach several workspaces to one metastore. They then see the same catalogs, tables and permissions.

  • Lineage and knowledge graph

    Lineage follows a table, down to its columns, up to its sources and down to what reads it. The knowledge graph maps the whole workspace and adds a business glossary and impact analysis.

  • Delta Sharing and marketplace

    Share Delta tables through an open Delta Sharing server, to recipients you create. The marketplace lists open datasets shared the same way.

Pipelines and compute

Load it, run it, promote it.

Bring data in, run work on a schedule and move it from development to production with a preview.

  • Ingestion

    Load files, folders, S3 compatible buckets, Google Cloud Storage, Azure storage, databases and REST APIs into tables. Incremental state means data that was already read is not read again.

  • Jobs and pipelines with retries

    A job is a graph of tasks: notebooks, Python, SQL, dbt, shell commands, other jobs and conditions. Failed tasks retry with backoff and jitter, and every attempt is on record.

  • Environments and promotion

    Keep Development, Staging and Production as isolated environments. Promote queries, notebooks, jobs and more between them with a preview and a rollback.

  • Serverless and cloud compute

    Connect cloud accounts and SSH hosts, launch instances and run work on serverless workers that start when work arrives and scale to zero when idle.

Data quality

Find out before your readers do.

Checks, monitors and contracts watch your tables, and personal data is found before it spreads.

  • Data quality checks

    Expectation suites run on demand, on a schedule or as a gate in a job. A failing suite stops the tasks that depend on it.

  • Anomaly detection

    Monitors collect table metrics on a schedule and flag points that look unusual against the table's own trusted history.

  • PII scanning and erasure

    Scan text columns for personal data. Erase a data subject from the tables that hold them and keep a certificate you can verify later.

  • Data contracts

    A contract states a table's columns, expectations and freshness promise. Drift is sorted into breaking and non-breaking, and a CI check can stop a deploy before it breaks readers.

Vyx and AI

An assistant that works inside your permissions.

Vyx answers questions about your data and fixes its own SQL. Every model it uses is one you connected.

  • Vyx assistant

    Ask in plain language and Vyx writes the SQL, asks before it runs anything, and runs it as you, so row filters and column masks still apply. A failed query is repaired automatically up to a limit, and your chat history is kept encrypted for you.

  • Agents, MCP and skills

    Build agents with tools, skills and MCP servers, evaluate them and deploy them. Lakehouse Studio is also an MCP server, and a client can do only what its token owner could do.

  • AI Search

    Create vector indexes over your tables and documents for agents, MCP and retrieval.

  • AI Gateway

    One gateway for 22 providers, from hosted services to Ollama, vLLM and LM Studio on your own hardware. Routes send each feature to a model, with weights and fallbacks.

Machine learning and documents

From experiment to endpoint to document.

Track the work, serve it, and turn files into checked data.

  • MLflow experiments

    Track runs, compare models and inspect agent traces logged with MLflow.

  • Models and serving

    Register models and put them behind serving endpoints that you manage from the same place.

  • Feature Store

    Manage feature views with Feast and serve features to your models.

  • Document Intelligence

    Parse files, extract the fields you define, classify them and decide what happens next with a policy. Every result has a confidence and a citation, goes to a review queue when a policy says so, and lands in a table. The same steps are SQL functions.

Databases, apps and tools

The rest of the workbench.

A real PostgreSQL server, apps next to your data, a terminal client and a place to learn.

  • Lakebase Postgres

    A managed PostgreSQL server with extensions, a REST Data API, backups with point-in-time recovery and a pooled connection endpoint.

  • Apps hosting

    Run Streamlit, FastAPI and Flask apps next to your data.

  • SDK and CLI

    The lakehouse command and a Python SDK cover queries, exports, jobs, dbt, tokens and roles, for scripts and CI.

  • Learn and Help

    Short hands-on courses, guided demos and product tours run on a sample warehouse, and the Help center ships with the app.

Security and trust

Built for teams that answer for their data.

Who may do what is decided in one place and enforced by the server. Where the product has limits, the documentation says so.

  • It runs on your servers

    Lakehouse Studio is installed by your team. Its data lives in a directory you control.

  • Vyx only uses models you connect

    Vyx talks only to the models you add. Mark each model as local or external, and a governance template can keep prompts on your network.

  • Honest about limits

    The security guide lists residual risks plainly, such as who can read keys and what a code runner can reach.

  • Account console

    Manage workspaces, shared metastores, people, usage and account-wide policies above the workspaces.

  • Roles and access reviews

    Built-in roles, presets and custom roles with a version history. A report lists who holds what, how, and when it was last used.

  • SAML and OpenID Connect

    Single sign-on with your identity provider, including Okta, Microsoft Entra ID, Google Workspace and Keycloak.

  • MFA and passkeys

    Authenticator apps, passkeys and single-use recovery codes. An account policy can require a second factor.

  • Token scopes

    API tokens carry scopes and never exceed what their owner may do.

  • Row filters and column masks

    Governance policies are enforced by rewriting every SQL statement, so they apply in the editor, in jobs and in Vyx.

  • AI encryption and governance templates

    Vyx chats, runs and traces are encrypted at rest per person. Templates decide which models may answer, what may enter a prompt and what Vyx may do.

  • Audit

    Every change and denial is written to an audit log that account admins read.

  • Backups and operations

    Documented backup and restore, an Operations page for health, and Lakebase backups with point-in-time recovery.

How it works

Three steps, no surprises.

  1. Step 1: Run one command on a new server

    On an Ubuntu 24.04 server you control, paste the command and enter your work email. The 30-day trial starts and the installer is checked before it runs.

  2. Step 2: Answer a few questions

    The installer asks for the address of the server, backups and who may reach SSH, then sets up the workspace, the firewall and HTTPS.

  3. Step 3: Create your owner account and start working

    Open the address, create the owner account and invite your team. Your data stays on your own server.

Free for 30 days

Install it on your own server.

Open a terminal on a new Ubuntu 24.04 server and paste one line. It asks for your work email, starts the trial and installs the workspace.

curl -fsSL https://forewarden.com/install | sudo bash

The command downloads the installer from this site and checks it against a fingerprint before it runs. It needs root, so it starts with sudo. You can read the script first at forewarden.com/install.

What your server needs

  • A new server running Ubuntu 24.04 LTS on x86_64, with nothing else on it. The smallest one we installed on has 2 CPUs and 4 GB of memory.
  • Root access with sudo, and ports 80 and 443 open to the internet.
  • A domain name, such as studio.yourcompany.com, that points at the server. It is used for the free HTTPS certificate.
  • A work email address for the trial license.

The trial, plainly

  • One trial for each email address, 30 days from the moment you start it.
  • We do not check that you own the address. A trial key is made for whoever asks, so use your own.
  • Your data never leaves your server. Only your email, the server name and a one-way server fingerprint reach us. The full list.
  • When it ends, the workspace locks. Nothing is deleted.

Open source inside

Built on work we are grateful for.

Lakehouse Studio runs on your own servers and stands on these open source projects. Thank you to the people who build and maintain them.

  • DuckDB
  • dbt Core
  • Delta Lake
  • MLflow
  • PostgreSQL
  • Apache Arrow
  • Apache Spark
  • LiteLLM
  • Docling
  • FastAPI
  • Next.js
  • React
  • Apache ECharts
  • Monaco Editor
  • React Flow
  • Great Expectations
  • Presidio
  • Feast
  • LanceDB

Questions

Before you ask.

Who is Forewarden?

Forewarden is the company behind Lakehouse Studio. Lakehouse Studio is a workspace for SQL, notebooks, dbt, jobs, dashboards and AI assistance on top of your own data.

How does the free trial work?

You run one command on your own Ubuntu 24.04 server and enter your work email. You get a license that is valid for 30 days. There is no card and no account to create here. The 30 days start when the trial is requested, not when the install finishes.

What happens to the email I enter?

It is used to make the license and to stop the same address from starting many trials. See the privacy page for exactly what is stored.

Where does Lakehouse Studio run?

On servers your organization controls. It uses open source parts, and its data stays in a directory you choose.

Does Vyx send my data to someone else?

Only to the model you connect. A model on your own network stays on your network. A hosted model receives the question, table and column names and masked sample rows, and a governance template can restrict that.

What happens after 30 days?

The workspace locks and shows that the license has ended. Nothing is deleted: your data stays on your server, and it opens again when a license is renewed.

Is Lakehouse Studio open source?

It is built on open source projects, thanked above. Lakehouse Studio itself is proprietary software.