Data work, end to end.

Often you should not move all of it, and we will say so. The first thing we do is work out what is straightforward, what is expensive, and what is not worth touching. Some workloads are cheaper and simpler where they are. What tends to be worth moving is the part where the jurisdiction matters: the tables your auditor asks about, the data you cannot describe to a regulator without a caveat. We will tell you which is which before there is a contract, and the answer is sometimes that you do not need us.

A lakehouse on an open table format, orchestrated pipelines, compute for the jobs, and whichever of dashboards, notebooks and model tracking your team needs. It runs in a tenant of your own, isolated from every other customer, on clusters dedicated to this platform. Every component above the storage is open source and you can read every line of it.

No, and not having one is the common case. If you have one, they get a working platform instead of a stack to source and assemble, and they run it. If you do not, we run it and your analysts get dashboards and notebooks without anyone on your side having to learn to operate the thing underneath them.

They sign in and open a notebook. Each person gets their own workspace, already wired to your compute and your model tracking, so there is nothing to install locally and no credentials to pass around. An analyst who only wants a chart opens the dashboards instead and never sees either.

Yes. You share a named table with that partner over an open protocol, so they read the live table from their own tools and never receive an export that is out of date by the time they open it.

We run the training and tracking side: CPU compute and the usual Python stack for the work, model tracking for recording what was trained, on which data, with which result. Inference and training on GPU clusters are not part of this platform. Our AI services cover them, on our own GPUs in Amsterdam: see the Artificial Intelligence page under Solutions.

Whichever you prefer. We can operate the platform, or hand it to your team and stay available. Either way, the engineers who designed it stay on it, so there is no handover to a support tier that has never seen your environment.

The orchestrator retries on the schedule the pipeline declares and records the failure with its logs, so the first question, which run broke and on what, is answered before anyone opens a ticket. Beyond that you are on the same support plans as the rest of what we operate: the free baseline on everything, and a paid tier if you want a faster answer. Same engineers, same phone number.

Yes, in full, and it is not a line item. This is a managed service: backing up the platform and the tables in it is our job rather than something you configure or buy separately. What we will not do is print a retention window on a web page, because the honest answer depends on what you are keeping and for how long you have to keep it. Ask on the call and you get the real one.

Yes. Sign-in is yours to control: federate the identity provider you already run, or use our IAM and SSO. Either way an account reaches your tenant and nothing else. Notebooks are one workspace per person. Dashboards carry a read-only role that can open what it is given without reaching the database list or the datasets behind it, which is the difference between letting somebody read a chart and letting them query the warehouse.

The discovery call, and it is free. 60 minutes, an engineer rather than a salesperson, and a written diagnosis inside a week: what we would build, in what order, and what we would leave alone. Plenty of people take that and do the work themselves. That is a fine outcome.

That depends on the state of your sources, which is exactly what the review stage exists to establish, so any number we gave you before looking would be made up. What we will commit to is telling you the shape of the answer in the written diagnosis rather than after a discovery phase you have paid for.

Not by the format. Your tables sit on ordinary object storage in an open format, which anything that speaks that format can read, and every component above them is open source. Moving a working data platform is never trivial and we are not going to pretend otherwise. What we will say is that nothing you build here ends up in a shape only we can open.

Madrid, Munich, Helsinki or Amsterdam, and you pick which at the start. It is our own object storage in our own datacenters, and the compute runs on our hardware in the same place, so the platform processes your tables in the site where they are stored. Your team reaches it over the public internet like any other service, authenticated before the request gets through.

DIVERSITY is a European company with no parent company, US or otherwise, so there is no entity above us for an order from outside the EU to be served on. That is a structural answer rather than a policy one: it does not depend on a setting you have to trust us to have left alone.

First, the honest part: the financial-sector rules do not require you to host in the EU, and anyone telling you otherwise is selling something. What they do require is that you can show you manage the risk of piling critical dependencies onto the same few providers, and that you have an exit plan you have actually tested. European supervisors now publish a list of the critical ICT providers the sector leans on hardest, and the large hyperscalers are on it. We are not. So moving a workload here reduces how much of it rests on the providers on that list.