Liminary uses AI to power features across your saved knowledge, including summarization, question answering, recall, synthesis, and analysis of text, images, and content from your documents, spreadsheets, and web pages.‍ Here's how we choose, run, and protect the models behind those features.

We benchmark, then pick We run our own benchmarks to determine which models are best in class for each task. If an open source model outperforms a commercial one on our benchmarks, or delivers a better user experience, we'll use it. Performance for you is the only criteria.

Your data stays yours We run all models in our own cloud environments and using APIs under enterprise agreements with model providers. Your content is never used to train any generative AI model. Cloud providers do not have access to your data.

You choose your chat model Liminary gives you the option to pick which model you want to use for chat. The model families we currently support include:

Anthropic Claude · OpenAI GPT · Google Gemini · Amazon Nova

We continuously evaluate new models and will add or swap them based on our benchmark results.

Google user data Liminary's use of information received from Google APIs adheres to the Google API Services User Data Policy, including the Limited Use requirements. We do not use Google user data to train, improve, or build any AI or machine learning models.

For more details, see our Privacy policy and Security policy‍