We do work with Cohere and know them well. We work with them on projects in the responsible AI area, for example, but quite frankly, you need to make available AI models from a variety of sources to have a robust approach.
The question is how you make them available to your workforce so that you're also responsible for where information goes and how you're employing those models. The other thing is that you have to add expertise to the use of the models, because they obviously have issues.