Menu
Book a demo View on GitLab →
Optimist · public proof-of-concept · seeking testers

Deploy it. Build it. Sail it.

An open, modular, Python-native data & AI platform you operate yourself. From a laptop to a fleet — on European infrastructure, without vendor lock-in.

Built on
dlt DuckDB DuckLake dbt Dagster Marimo Elementary

Optimist — a complete data & AI platform on your laptop.

Named after the sailboat designed for learning, Optimist is a complete, self-hostable data platform that integrates best-in-class open-source tools into a coherent Lakehouse architecture. From ingestion to insight, with strict modularity at every layer.

01

The full data journey, integrated

From raw data ingestion (dlt) through storage (DuckDB + DuckLake) and transformation (dbt) to orchestration (Dagster) and exploration (Marimo). One Python install, one consistent platform.

02

Crow's Nest

A unified monitoring and control interface across every integrated tool. No five tabs to keep open, no scattered dashboards — one place to operate the platform.

03

Strictly modular, no lock-in

Every layer is decoupled. Swap any tool — ingestion, storage, transformation, BI — without rebuilding the system. Long-term flexibility by design, not by promise.

04

Self-hosted by default

Runs where you want it — laptop, VPS, European cloud, on-premises. Your data never leaves your infrastructure unless you decide so.

05

AI-ready by design

Your data in DuckDB enriched with Elementary context layer is the ideal substrate for LLM queries, RAG pipelines, and AI agents. The same data that feeds your dashboards feeds your AI layer — clean, governed, and always on infrastructure you control.

~/datavloot
# 1. Install the Optimist
$ pip install datavloot[optimist]
→ resolving dependencies...
✓ installed
 
# 2. Create your project or install the demo
$ datavloot new / datavloot demo
✓ ready to launch
 
# 3. Launch the Crow's Nest
$ datavloot start
✓ Optimist underway → localhost:8080
$

Scaling up is a config change, not a migration.

Build your data pipelines on the Optimist locally. When you're ready for team scale, switch one flag — the same pipelines, the same dbt models, the same orchestration logic — and the Falcon takes over on your infrastructure of choice: EU cloud, self-hosted, or hybrid.

Optimist
on your laptop
# datavloot.yml
vessel: optimist
storage:
  type: duckdb
  path: ./data/local.db
orchestration:
  mode: local
$datavloot start
Falcon
on your infrastructure
# datavloot.yml
vessel: falcon
storage:
  type: iceberg
  # cloud:
  endpoint: eu-central.hetzner.cloud
  # or self-hosted:
  # endpoint: minio.your-server.local
  bucket: datavloot
orchestration:
  mode: cloud
$datavloot deploy
Same pipelines. Same dbt models. Same orchestration logic. What changes is where it runs and where your data lives — not how you wrote it. No re-platforming, no rewriting, no rebuild.

No lock-in

Open standards, open licences. Take your data and pipelines anywhere, anytime.

Strictly modular

Swap any tool — ingestion, storage, transformation — without rebuilding the stack.

Sovereign by default

You decide where your data lives — laptop, on-premises, European cloud, hybrid.

Grows with you

Start local in Python. Move to Docker. Adopt Kubernetes when, and only if, you need to.

Four vessels, one course.

Optimist is the starting point. As your team and data volume grow, Datavloot scales along — to larger, more capable configurations without starting over. Same architecture, more capability.

Tier 02 · Falcon Roadmap

Falcon

For the small team — production-grade.

Everything Optimist offers, plus real lakehouse storage with Iceberg on S3 and shared dashboards via Marimo apps.

Runs on: Docker Compose + S3-compatible storage
Cloud, hybrid, or fully self-hosted. EU providers (Hetzner, Scaleway, R2, Exoscale) for managed; MinIO or Garage for on-premises.
Optimist + Apache Iceberg
Tier 03 · Catamaran Future

Catamaran

For growing teams — governance and scale.

When teams grow, governance follows. Adds federated querying with Trino, policy enforcement via OPA, and metadata management through OpenMetadata.

Runs on: Kubernetes
Falcon + Trino OPA OpenMetadata
Tier 04 · Yacht Future

Yacht

For enterprise — in development.

Full enterprise configuration with identity management at scale, AI-powered data agents, and data sharing across teams. In development with early adopters.

Runs on: Kubernetes + Helm + GitOps
Catamaran + Keycloak AI agents

Your data. AI-ready.

A governed data pipeline is the ideal foundation for LLM-powered applications. Datavloot connects your data layer directly to your AI layer — on infrastructure you control, with models you choose.

Clean data as AI context

Every layer in Datavloot — ingestion, transformation, quality checks — produces data that's ready to serve as context for LLMs. The same DuckDB warehouse powering your dashboards powers your AI queries and RAG pipelines.

Private inference by default

Run open-weight models on your own infrastructure. No API keys, no data sent to third-party clouds. The same sovereignty principle that governs your data pipeline governs your AI layer.

Modular AI, no lock-in

Connect any LLM framework to your data layer. Swap models without rebuilding pipelines. The same modularity that makes the data stack replaceable makes the AI components replaceable too.

Captain & crew

Build your platform with AI

The Optimist is designed to be built with an AI assistant from day one — not as an abstraction that hides what you're building, but as a way to move faster while you learn it.

A dedicated agent skill ships inside every project created with datavloot new. Hand it to your AI assistant at the start of a session so it works within the toolkit's conventions.

You Direct the session — your domain knowledge, your data sources, your output requirements.
AI Handle implementation — dbt models, Dagster assets, dlt pipelines — within the toolkit's conventions.
You Review, question, approve. The understanding stays yours; the AI accelerates getting there.

Optimist is looking for early sailors.

A working proof-of-concept runs today on public AIS shipping data. The next step is to run it on your data, with your use case — and help shape what Optimist becomes.

Early involvement

You use Optimist before public release and directly influence what gets built next.

Direct line — no ticket queues

Speak directly with the team behind the platform. Email, call, message — your questions get priority.

A working solution for your case

Together we map your sources, transformations, and outputs — and build a working Optimist around them.

Mutual, no obligation

No costs, no contracts. A learning round on both sides — with an honest expectation of openness and feedback.

Sound good?

Send a short email with what you do, what data you work with, and what you hope to learn. We respond within a few days.

Apply as tester →
info@datavloot.nl

Three steps to your first pipeline.

From install to a working data pipeline — usually within an afternoon.

01 · INSTALL

pip install

Optimist installs as a Python package. One command, one virtual environment. No infrastructure prerequisites.

02 · CONNECT

Wire up your sources

Use dlt to connect to APIs, databases, or files. 60+ pre-built connectors available out of the box.

03 · TRANSFORM

Build your models

Write SQL models in dbt, orchestrate with Dagster, explore in Marimo. The Crow's Nest brings everything together.

The tools that power Datavloot.

Every tool in the stack was chosen for its modularity, openness, and Python-native fit. Together they form one coherent platform — each component replaceable, none mandatory.

Ingestion

A Python library for building data ingestion pipelines. Connects to 60+ pre-built sources — APIs, databases, files — with automatic schema inference and incremental loading.

Compute

An in-process analytical database — SQLite-style simplicity for analytics workloads. Runs inside Python with no server to manage, processes millions of rows fast on a laptop.

Storage format

An open lakehouse table format on top of Parquet. Brings warehouse features — ACID transactions, schema evolution, time travel — and a technical catalogue for your data.

Transformation

A SQL-based transformation framework. Write modular SELECT statements; dbt handles dependencies, materialisation, testing, and documentation. Raw tables in, clean models out.

Dagster dagster.io →
Orchestration

An orchestration platform for data pipelines. Define, schedule, and monitor pipelines as Python code, with a local UI for observability and debugging.

Marimo marimo.io →
Exploration

A reactive Python notebook. Query your data in SQL or Python with built-in UI components — sliders, dropdowns, tables — for exploration and ad-hoc analysis.

Observability

A dbt-native data observability tool that allows you to monitor pipeline performance, data quality and a rich context layer.

Beyond Optimist: Falcon adds Apache Iceberg (lakehouse storage), with Marimo extended for shared dashboards. Catamaran adds Trino, OPA, and OpenMetadata. Full documentation per tier is in development on GitLab.

Which vessel fits your situation?

Five questions about your team, data, and operational context. The result is the configuration that matches your profile today.

Why Datavloot — and who's behind it.

A modular, sovereign data & AI platform built from open-source tools, by people who run it themselves.

What Datavloot stands for

The current generation of data platforms — Databricks, Fabric, Snowflake — all run on US hyperscaler infrastructure, even when the data sits in European data centres. That's a sovereignty gap that grows with every new regulation, every geopolitical shift, every CLOUD Act reminder.

Datavloot's bet: a modular, open-source data stack you run wherever you want. No subscriptions, no vendor lock-in, no Cloud Act exposure. Built from best-in-class open tools, integrated into a coherent platform, available under truly open licences — not open-core wrapped in marketing.

The aim is to give European data teams a real alternative, and to contribute one reusable, trustworthy building block to the digital commons.

About Stan

Stan Verwer
Stan Verwer
Founder, Datavloot
Data solution architect

Biomedical Engineer (MSc) turned data engineer — ten years across cloud and on-premises platforms, mostly in healthcare. Currently a data solution architect at the Cultural Heritage Agency (RCE), part of the Dutch Ministry of Education, Culture & Science.

That work surfaced a gap I couldn't unsee. The modern data tooling I knew best — Microsoft/Azure, Snowflake, Databricks — couldn't meet the sovereignty bar that public-sector projects increasingly require. Existing open alternatives demanded Kubernetes from day one, started at enterprise sales cycles, or turned out to be open-core with the useful parts behind a paywall. Datavloot is my attempt to close that gap.

About Eva

Eva van Kalmthout
Eva van Kalmthout
Co-creator, Datavloot
Data platform engineer

Data engineering has quietly become about big-tech certifications — Databricks certified, Azure certified, Snowflake certified... As if you need a new driver's licence for every car brand. That bothered me a little.

The question that stuck: what if we brought it back to the basics? Not learning to use one of the big tech platforms, but learning to build and own one yourself — from open-source tooling that's replaceable when it no longer fits your needs. This also provides us engineers with the self-reliance to build towards any requirement we have, including souvereignity. Power to the builders!

Guidance from someone who builds it.

Optimist is free and open. For organisations that need a faster ramp-up, a tailored implementation, or strategic guidance, we offer paid services — directly from the team building the platform, no overhead.

Schedule a call →

Implementation

Set up Datavloot on your infrastructure, wire it to your data sources, and get your team operating it independently. Project-based, fixed scope.

Advisory

Architecture review, tool selection, migration planning. Independent, vendor-neutral. Half-day sessions or longer engagements.

Support & maintenance

Ongoing operational support — upgrades, troubleshooting, platform evolution. Monthly retainer with priority response.

Book a demo

Let us know when you're available and we'll schedule a 30-minute call — together we'll explore whether Datavloot is the right fit for your situation.

You'll receive a confirmation at the email address provided.

Request received

We'll be in touch as soon as possible to schedule a time.