Home › On your terms
On your terms: what we commit to in every topic
“Intelligent Enterprise, on your terms” stands above everything we build. It comes with four commitments that hold in each of our three topics: Agentic Commerce, Process Automation and Agentic Engineering. This page sets out what each commitment means and where your data goes, depending on where your AI runs.
You decide where your data goes
AI models need your data to work. The data goes where the model runs, and you decide where that is. With it, you also decide whether your data stays with you or can end up with organisations and authorities you have no influence over.
We distinguish three places.
Cloud. For us, Cloud means models from third-party providers, often in the US. Your data goes to that provider. Neither you nor we have any influence over how it handles the GDPR or over requests from authorities. The strongest models run there, and it is the fastest start. The Cloud suits public data.
In-House Cloud. An In-House Cloud means servers only for you, hosted at a provider of your choice or in your own data centre, down to a single machine for one user. Your data stays on these servers. Only those you admit have access, we under a data processing agreement. We run the servers like a cloud, and operations are part of the package. There is no token bill. Your own account with a cloud provider counts too, as long as the servers in it run only for you. If you use the provider's model service there, such as Azure OpenAI or Amazon Bedrock, your data goes to third-party models, and that is Cloud.
Hybrid. Hybrid is a special case: a small model in your In-House Cloud checks each request and pseudonymises it through our Privacy Gateway. Only placeholders go to the cloud model; the original with all its values stays with you. Which data you handle that way is your call, because only you know the risk if pseudonymisation is ever incomplete. For many, Hybrid is the way in, because they do not have to buy hardware for local AI first.
Which data goes to which place follows the data class, and how you classify a document is your organisation's decision. Typically only public data goes to the Cloud. Everything else stays in your In-House Cloud or goes out hybrid, as placeholders only. Documents under NDA might then sit on hosted servers, and strictly confidential material on your own premises.
One project can use several places. For a quotation, for example, the tender text goes to the Cloud pseudonymised, while the costing and the margins are seen only by models in your In-House Cloud. The more clearly a task is described, the smaller the model can be, and the more likely it is to run with you. You can change your choice later.
The platform with storage, knowledge base and workflows runs on our servers under a data processing agreement, or entirely in your In-House Cloud.
The solution belongs to you
What we build for you belongs to you. We build on open standards and an open-source base, without lock-in; automation, for example, runs on Liquid.MX, under which we run open-source integration tools. The platform brings sign-in, roles, backup, operations and the path from development to production. We build your business logic and your interface.
Where the solution runs is a decision you make once, as described above. Separately, you decide who administers it. Either way it is operated like a cloud, and we look after servers and updates. There are two ways to administer it:
- As a service: we make all the changes the platform needs on your behalf, such as setting up users and granting rights.
- Self-service: you get a portal and administer your solution through it yourself.
Your team develops the solution further and can take over administration at any time. What we build with is described in our toolbox.
Value first, then you decide
Every topic starts with a first step at no charge. In process automation, for example, that is the process check: you pick one process that runs by hand today, upload three typical documents and get one page back showing what can run by itself. After that you decide stage by stage, and you know scope and price before each stage.
Measured before and after
At the start we agree with you what counts as success, and at the end we measure it. In process automation that might be the hours per week your team spends typing documents in today. In Agentic Commerce it is retrievals, citations and price parity across all channels. For AI visibility we state results as a range and give no forecast of the impact.
What does not leave your premises
In Agentic Commerce, agents work with purchase prices, terms and price floors. These agents run on models with you. Purchase prices, terms, customer prices, margins, stock, orders and revenue do not leave. For cloud models only texts with placeholders go out; the original stays with the Privacy Gateway on your premises.