The terms we use, defined plainly. Each one is a question we get asked in the first conversation.
Sovereign AI is artificial intelligence that runs entirely on infrastructure an organisation owns or controls, inside its own jurisdiction, with no third party in the data path. The models, the data, and the compute all stay on servers the company governs, so sensitive information never leaves the building and is never sent to an external AI vendor. For European companies it usually also means the compute runs on EU soil, under EU law.
A purpose-built agent team is a set of AI agents, each built for one specific job in a company’s workflow and tuned to that company’s data, deployed together as a team. It is the opposite of a generic, configurable platform: instead of buying a toolkit and modelling your work into it, you receive the specific agents your workflow needs, where each agent is both the logic and the interface. You deploy the agents you need, not a platform you have to configure.
An agentic AI system is software in which AI agents take actions to complete multi-step work, rather than only answering questions. An agent can plan, call tools, read and write data, coordinate with other agents, and carry a task through to a finished result. The shift from a chatbot to an agentic system is the shift from "it tells you" to "it does it".
Ticket-to-PR is a way of delivering software in which a work ticket (a described task, bug, or feature) is turned into a reviewed pull request automatically by agents. A person describes what is needed; the agents produce the code change, run it through review, and hand back a pull request ready to merge. It compresses the path from request to working software.
Small open-source models are language models with openly available weights, compact enough to run on a single company’s own hardware. Compared with large frontier models accessed over an API, they are cheaper to run, can run fully offline on your own servers, and keep data in your control. For many business workflows a well-chosen small model, tuned to the task, is accurate enough while being sovereign and far less expensive per use.
AI augmentation means adding agentic capacity on top of the expertise and operations a business already has, so people do more, rather than replacing them. Replacement-style AI removes humans from a task entirely. Cultivator AI builds augmentation: the agents handle scale, speed, and repetition, while people keep judgement, domain expertise, and the decisions that need them.
Tell us one workflow you want to give agentic capacity to and we will show you what a purpose-built team for it looks like.
Book a call