Frequently askedquestions
The platform
What exactly is datavloot?
A complete data & AI platform, assembled from proven open-source components for ingesting, storing, transforming, testing, analysing and orchestrating data. We bring together components that run in production all over the world, and make sure they work and can be operated as one whole. The source code of that assembly is free and available to everyone.
Is this software we buy, or a service?
Neither, in the usual sense. The software is open source and free — you can install it today without ever speaking to us. What you buy from us is guidance: setting the platform up on your sources and your data model, and passing on the knowledge so your people can run it themselves. No licence and no subscription is needed for the software, only an optional support contract.
Which components are in it?
Our first platform, the Optimist, contains the following components: dlt for ingestion, DuckDB with DuckLake for storage, dbt with Elementary for transformation and data quality, Marimo for analysis and AI, and Dagster for orchestration. Every one of them a project with a large community and worldwide adoption. The Crows Nest is a control interface (a cockpit) that we built on top. Free and open source as well, of course.
What if one of those components disappears?
Then you replace it. Each part performs one specific function of the platform and is decoupled from the rest. You can replace the storage layer or the visualisation layer without rebuilding everything else. With a closed platform, that is exactly the problem you cannot solve.
Delivery
How quickly is the platform rigged and ready?
The Optimist is on your infrastructure within a day. An implementation with your own sources and data model (a minimum viable product) takes ten working days on average: installation is one day, the rest goes into connecting your sources and modelling your data. In that process we turn your data into insight for the organisation together.
What exactly does the implementation cover?
That differs per organisation and we set it out in the quote. The core is always the same: we connect your sources, build a curated layer around one concrete question, and pass the knowledge on to your own people so the platform is theirs afterwards. Additional work is billed in half-days at the day rate, always agreed in advance.
Where will the platform live?
On your infrastructure and in your account. Preferably on what you already have: an existing VM environment, your own server, or your current cloud. Have nothing yet, or want to move to a European sovereign server? Then we set up a server together at a European provider. We host nothing and pass nothing on. Infrastructure rental is your own cost and is not included in the implementation.
Do we have to move to European infrastructure?
That depends on the sovereignty requirements of, or for, your organisation. If you are already on Azure or AWS, a Linux server runs there perfectly well. The gain is in open source code, no lock-in, and data you can take with you: the exit strategy. You can always move later.
What do we need to have ready before you start?
A Linux server with Docker and Docker Compose, 4 to 8 vCPUs and 16 to 32 GB of memory, disk space of three to five times the raw data, outbound internet to the source systems and inbound only port 443, a DNS name with a TLS certificate, and an administrator with root rights who is reachable during the project. If that is ready on day one, no day is lost to waiting. We can of course help with this, or look at your environment with you beforehand.
Does our data have to go anywhere?
No. Everything runs where you want it: your own machine, your own server, a European provider or a private cloud. Nothing goes to us, and nothing goes to a hyperscaler (Azure, AWS, Google) that you have not chosen yourself.
Support & maintenance
Who runs the platform after handover?
You do, in principle — that is our starting point: we pass on the knowledge so your organisation can run it itself. If you want to keep us around, you can; for €500 a month we look over your shoulder and make sure the platform keeps running and stays up to date. Further development is charged separately.
Do you offer an SLA or ongoing support?
An SLA with 24/7 availability and an on-call rota, no. Ongoing support, yes: for €500 a month we keep an eye on the runs, carry out the updates and reply within one working day. Cancel any month — and if you do, the platform simply keeps running.
Do we need a data professional of our own?
For day-to-day use, no: dashboards, analyses and exports can be used by people across the organisation without one. For maintenance and further development, yes. We offer documentation and a free online workshop for that, as well as paid training. See our services page for more. If you have no data professional on staff, we can fill that role ourselves for a while, or help you find a suitable freelancer.
What should that data professional be able to do?
SQL matters most: the models that turn raw sources into reliable numbers are written in dbt, and that is SQL with some structure around it. Some Python helps for connecting a new source with dlt. As far as we are concerned, SQL and Python are both basic skills every data professional should have. Basic Linux and Docker knowledge — enough to restart a container and read a log — is a plus. Version control with Git is needed to keep track of every change to the platform and to let several people work on it. These are also the areas we spend most time on during the knowledge transfer.
What is involved in keeping the datavloot platform running?
The Crows Nest (the control interface) shows on one screen whether the scheduled runs succeeded and whether the data quality tests are green; you restart a failed run from there. Updates need doing now and then, so the platform stays current and keeps working. Beyond that there is occasional work: connecting a new source, extending a model, or adding new users to the platform.
What if datavloot stops tomorrow?
Then your platform simply keeps running, because nothing of it sits with us. The source code is open and it runs on your infrastructure. And because it is open source, any organisation can pick it up and carry on.
What about updates?
The open-source components (dbt, DuckDB and so on) are maintained by their own communities and follow their own release rhythm. We make sure those changes land in the datavloot platform, so we release updates regularly that keep everything working as a coherent whole. During the knowledge transfer we show you how to apply them and what to watch for. You do the updating yourself, unless you take a support contract.
How often is the data refreshed?
On the schedule you set yourself: daily, hourly, or even near real time if the sources allow it. Users of the platform — through reports or AI agents, for instance — see the data from the most recent refresh.
Security
How do we know the software is secure?
Because you can look inside it. The datavloot source code is public, and the underlying components are projects with large communities, a public issue process and a published vulnerability policy. You can run your own scanning against it and have your own security team review it — with a closed platform you cannot.
Who can reach the dashboards and the data?
You arrange that with what you already have. Access to the environment goes through your own SSO, VPN or reverse proxy — the same route as your other internal applications. If you want authorisation down to the user, role or dataset within the platform itself, that is the next configuration on the roadmap.
What about backup and recovery?
The full state and code of the platform — data, models, configuration and history — lives in one folder and is stored safely in a git repository. Backup, retention periods and the like for the data itself can be set up at your chosen cloud provider, by you or by us.
Can we show where a number comes from?
Yes. Every transformation is captured in code and tested, and the origin of every number can be traced back to its source (lineage), for audits if needed. Elementary also runs quality checks on the fly, which you configure yourself, and stores the results in reports.
Does data go to an AI model outside our environment?
Only if you set it up that way yourself. The platform captures meaning, coherence and provenance so AI agents can make sensible use of it; which model you put on top, and where it runs, is your choice. There is no model in it that sends data out by default.
Scale & growth
What kind of organisations is the Optimist meant for?
For organisations with one data professional running the data platform, or a small team that does not need to build on it simultaneously. What kind of organisation it is, and how many staff it has — and so how many people open a report at the same time — does not really matter.
What if we outgrow the Optimist?
Then you move up to the Falcon: shared storage with a catalogue, colleagues working on the models at the same time, and access policy per user. The same components and the same way of working. The Falcon is on the roadmap for Q4 '26, so it is not here yet.
Pricing & terms
What does it cost?
The software costs nothing — it is open source and will stay that way. Guidance has fixed rates: a discovery day at €1,750, a tailored implementation at €7,500 as a pioneer rate, in-company training at €1,750 per day for up to eight participants, the open build-it-yourself training at €595 per person from three participants, and ongoing support at €500 per month. All amounts exclude VAT.
Why such a low rate?
As we are still a young start-up, we currently charge a pioneer rate.
Is the discovery day deducted?
Yes. If you go ahead with an implementation within three months, the cost of the discovery day comes off it. If you do not go ahead, you keep whatever was built and written down that day — it is yours, even if you buy nothing further from us.
Are there mandatory follow-up engagements?
No. No subscription, no minimum purchase, no renewal clause. Our aim is that you need us as little as possible afterwards.
What happens if the project overruns?
Overrunning because of our own estimate is our risk; the pioneer rate is a fixed amount. If it overruns because the scope grows — a fourth source, a second data model — we discuss that in advance and bill it in half-days at the day rate.
Get in touch.
is your question not here?
Call or email, or book a free 30-minute demo.