Platform
From source to dashboard. All of it, or just what you need.
Allodion is a managed data platform for European organisations, built entirely on open-source tools. Use the full stack, hosted by us, or slot Allodion into the storage and BI tools you already have.
Architecture
Built from open-source tools you already trust.
No proprietary core. Every layer is a proven open-source project, integrated, secured and run for you. If you ever leave, you take the same tools and formats with you.
Your sources
Databases · files · APIs · SaaS apps
Ingest & schedule
Develop & transform
Compute
Catalog
Storage
BI & dashboards
Across every layer
Identity, access & secrets
Zitadel · single sign-onOpenFGA · data permissionsOpenBao · secretsThe tools
What each part does, and why it is there.
Storage format
Apache Iceberg on S3
The open table standard for the lakehouse: transactions, schema changes and time travel on plain object storage. Readable by almost every engine.
Catalog
Lakekeeper
Keeps track of every table and who may touch it. One open REST catalog, so every tool sees the same, governed data.
Compute
DuckDB
Fast, efficient analytics for jobs and notebooks. Most datasets don't need a cluster, so you don't pay for one.
Compute at scale
Trino
A shared, distributed SQL engine for large datasets, many concurrent users and your BI tools. It also keeps your tables tidy and fast.
Ingestion
dlt
Loads data from APIs, databases and files in a few lines of Python, with incremental loads and schema changes handled for you.
Transformation
dbt-core
Turns raw data into trusted models with plain SQL, tested, documented and built in the right order. The same code runs on DuckDB and Trino.
Orchestration
Apache Airflow
Schedules and monitors your pipelines, with retries and alerts. Each client gets its own isolated instance.
Notebooks
JupyterHub
Python and SQL notebooks for exploration and development, signed in with your single sign-on account.
BI & dashboards
Apache Superset
Dashboards and SQL exploration for the whole organisation.
Identity
Zitadel
One login for every tool, with multi-factor authentication. Connect your existing identity provider.
Permissions
OpenFGA
Fine-grained access down to schema and table level. Finance sees finance data, HR sees HR data.
Secrets
OpenBao
Your source credentials and API keys, stored encrypted and handed out only to the jobs that need them. Every access is logged.
A stack that keeps improving
The open-source data ecosystem moves fast, and so do we. We keep a close eye on new projects, and when a better tool comes along we may change our stack. Because your data lives in open formats like Apache Iceberg, a change of tooling never means a change of data. We handle the migration, announce changes well in advance, and keep your code working.
Hosted or your own
Slot in where you need us.
Every layer except processing can be hosted by Allodion or stay with you. Mix and match, and change your mind later.
| Layer | Hosted by Allodion | Your own |
|---|---|---|
| Sources | Not applicable | Always yours: databases, files, APIs and SaaS apps |
| Storage | Managed EU storage, with backups and scaling handled | Your own storage buckets, in a location you choose |
| Processing | Python and SQL notebooks and scheduled jobs, always run by Allodion | Not applicable |
| BI and analytics | Hosted dashboards and reporting | Connect the BI tools you already use |
AI
AI on your data, without sending it abroad.
Your lakehouse is the best foundation for AI: clean, governed and in one place.
Ask questions in plain language
Turn a business question into SQL in your notebooks and dashboards, so more people can work with the data.
Turn documents into data
Extract fields from invoices, contracts and emails in your pipelines, and land them as tables next to the rest of your data.
Enrich in your models
Classify, summarise or score text as a step in your transformations, tested and scheduled like any other model.
Chat with your data, safely
Assistants that answer from your own lakehouse, and only from the data each user is already allowed to see.
European models first
Built with Mistral in mind
Mistral AI is a French company, and one of the few frontier model makers based in Europe. Its open-weight models can run on European infrastructure, inside the same environment as your data.
Prefer another model? Bring your own. As with the rest of the platform, you are never tied to one vendor.
Included in every setup
The parts you should not have to build yourself.
Governance and single sign-on
Sign in with your existing identity provider. Roles and permissions apply across every layer you run with us.
Access management
Users, groups and permissions managed for you, so the right people see the right data.
Operations
Infrastructure, scaling, upgrades and monitoring of every tool are handled by Allodion, from the EU.
Support
Direct access to a European team of data engineers who know the platform inside out.
Identity
Keep the login your team already uses.
Your team signs in to Allodion with the accounts they already have. Accounts are created automatically on first login.
Login only, never your data. Your identity provider only confirms who someone is. Permissions, data and processing stay on Allodion, in the EU. And when you are ready to leave Microsoft, Google or Amazon behind, your users can move to our European identity service.
Pricing
Pay for what you use, or reserve and save.
Flexible
Pay as you go
You pay for the compute and storage you actually use. No commitment, so you can start small and grow when you need to.
- Billed monthly on usage
- Scale up or down at any time
- Ideal for pilots and changing workloads
Predictable
Reserved capacity
Commit to a fixed amount of capacity for a longer period, in exchange for a lower rate and a budget you can plan around.
- Lower rate than pay as you go
- Fixed, predictable monthly cost
- Ideal for stable production workloads
Either way, you see your usage and costs per team and workload in your monitoring overview. No surprises at the end of the month, and no switching or exit fees.
Onboarding
From first call to production.
01
Intake
We map your sources, storage and BI landscape, and decide together which layers Allodion runs.
02
Connect
Your sources and, if you choose, your own storage and BI tools are connected. Single sign-on is set up.
03
Build
Your team builds pipelines and models in Python and SQL notebooks, with our support.
04
Go live
Jobs are scheduled, access is governed and your organisation works from one platform.
Coming from Databricks, Snowflake or Fabric?
The concepts carry over. Here is where you will find them on Allodion. What changes is who runs it, under which law, and that every part is open source.
| Concept | What you use today | On Allodion |
|---|---|---|
| Table format | Delta Lake / OneLake | Apache Iceberg |
| Catalog & governance | Unity Catalog / Purview | Lakekeeper + OpenFGA |
| SQL warehouse | SQL Warehouse / Fabric Warehouse | Trino |
| Notebooks | Databricks / Fabric notebooks | JupyterHub |
| Ingestion | Data Factory / pipelines | dlt |
| Scheduling | Workflows / Data Factory | Apache Airflow |
| Transformation | dbt on any of them | dbt-core |
| Dashboards | Power BI / AI/BI | Apache Superset, or keep your own |
| Identity | Entra ID | Zitadel, federated with your IdP |
Join the early access programme.
We are onboarding a small group of European teams.