Skip to Content

Artificial intelligence

AI model configuration and training

We install AI models for specific uses and assess whether they need adapting with your data. We start with the task the model needs to perform and how you will check the result.

A model for a specific task

Tell us what the model will receive and what you expect it to produce. For example, you might want to sort support requests by topic or extract fields from a set of documents. A few representative examples help us assess the language, accuracy and resources the task requires.

Model installation and adaptation

We install the model and set up the environment where it will run. If the task calls for a change in model behaviour, we can assess training on a set of reviewed examples. For a project that needs to look up documents, we can also consider giving the model access to relevant documents that you can update as the information changes.

Tests with representative examples

Keep a few examples aside to check the result at the end of testing. Tell us which errors would be critical and who would review them. If the project uses internal documents or personal data, explain any access and processing restrictions before sharing samples.

Tell us what the model should do