Gradient AI, a startup where developers can build and customise AI applications in the cloud using large language models, announced its partnership with database giant MongoDB. The partnership is said to strengthen the collaborative ties between the two companies helping Gradient AI users to build applications using the data platform. 

Developing custom AI solutions that leverage both fine-tuning and Retrieval Augmented Generation (RAG) can pose challenges. However, individuals can build robust and cost-effective solutions with the assistance of Gradient and top frameworks such as MongoDB and LlamaIndex. 

Chatbot With Multi-Integration

Financial chatbots can be created with MongoDB, where RAG can be used to reduce hallucination, MongoDB for Vector Store and LlamaIndex for RAG Orchestration. For instance, LlamaIndex can be used for uploading a PSF (point string function) into a raw string. Through Gradient’s web API’s, the models can be fine tuned, generate completions and embeddings from one platform. Connections can be made with Atlas Vector Search, and LlamaIndex can be used to bring together all the elements. 

Source: GradientAI

Last month, Gradient came out of stealth mode with a $10 million funding led by Wing VC, Mango Capital, Tokyo Black and other VC platforms. The company was started with the focus to help enterprises unlock the full benefits of language models which require a reliable way to add private, proprietary data to them. 

Gradient users are not required to initiate the training of LLMs from the beginning as the platform provides a selection of open-source LLMs, including Meta’s Llama 2, that users can fine-tune according to their requirements. Furthermore, Gradient offers pre-trained models designed for specific use cases such as data reconciliation, context gathering, and others that cater to various industries.