Datasaur's LLM Lab provides a starting point for companies building custom AI models that aren't met by commercial, cloud-hosted neural networks. The tool can be used to train custom neural networks on organizations' internal data, reduce infrastructure costs associated with running an AI model, and compare different foundation models. LLM Lab supports foundation models like Llama 2 and technologies like Pinecone, a vector database designed to store neural networks' data.
Key takeaways:
- Credal AI Inc. and Datasaur Inc., two AI startups, are focusing on making it simpler for enterprises to build secure AI software.
- Credal has raised $4.8 million in seed funding for its platform, which helps companies protect the input data they enter into AI models. The company currently has 11 customers and plans to use the funding to expand its base, hire more employees and build new features.
- Datasaur debuted a new tool called the LLM Lab, which can be used by companies building custom AI models for cybersecurity reasons or because their requirements aren’t met by off-the-shelf models. The tool aims to speed up development and can be used to train custom neural networks on organizations’ internal data.
- Both companies' platforms also simplify other AI-related tasks, such as routing user requests to the most suitable model, tracking neural network usage and costs, and reducing infrastructure costs associated with running an AI model.