Saige AI leverages a large language model (LLM) that has been tuned by the Dream See Do engineering team for our platform and your discreet needs. It streamlines course creation by generating suggested course layouts, content blocks, response options, quizzes, surveys, and assessments. Learners also benefit from instant feedback provided by Saige AI's virtual coaching feature, which has been trained on coaching principles that are aimed to help learners develop insights and take action, without giving advice.

Saige AI utilizes Anthropic's Claude foundational models, through their API. We chose to partner with Anthropic specifically because they continually choose to focus on AI safety, and are continually doing cutting edge research to ensure the output of their models is safe and not harmful to users. You can read more about their stance here.
​
​

LLMs are trained on very large amounts of data, and during the training develop relationships between words and concepts, similar to how the brain does. It then uses that understanding of language to generate coherent and contextually appropriate responses. Under the hood it is using mathematical probabilities to choose the next word based on what it has already generated. In this way, it doesn't just pull text from its training data, but rather it uses its understanding of language and the world to generate completely new material.

You can safely include intellectual property content when generating courses with Saige AI. We store the data securely on our servers, and ensure that it is never used to train any AI models. Anthropic explicitly does not train any IP or material when using their API, here is their documentation. It is only used during the generation phase.
​
​

If you want to dive further, you can view the LLM terms of service of the model here.