Databricks today announced several innovations to its Mosaic AI platform to help customers build production-quality generative AI applications. The company is investing in three key areas: support for building compound AI systems, capabilities to improve model quality, and new AI governance tools. 

Organisations are struggling to transition generative AI projects from pilot to full-scale production due to privacy, quality, and cost concerns. While foundation models have significantly improved, they still face challenges in producing consistently high-quality results. To address these issues, organisations are moving beyond deploying a single large model and instead adopting compound AI systems. 

This approach utilises multiple components, including various models, retrievers, vector databases, and tools for evaluation, monitoring, security, and governance, resulting in higher production quality and more accurate, safe, and governed AI applications.

“We believe that compound AI systems will be the best way to maximise the quality, reliability, and measurement of AI applications going forward, and may be one of the most important trends in AI in 2024,” said Matei Zaharia, co-founder and CTO at Databricks. “Databricks is uniquely positioned to capitalise on these trends with the investments we’re making to improve quality, augmenting the model with real-time data and agents and tools to give it new capabilities it has little knowledge of.”

To support customers in building production-quality generative AI applications, Databricks is launching several new features:

Mosaic AI Agent Framework and Mosaic AI Tools Catalog help organisations build compound AI systems. The Agent Framework enables developers to quickly and safely build high-quality RAG (Retrieval-Augmented Generation) applications using foundation models and enterprise data. The Tools Catalog allows organisations to govern, share, and register tools using Databricks Unity Catalog, ensuring secure and governed use of tool-enabled models.

Mosaic AI Quality Lab is an AI-assisted evaluation tool that automatically determines if outputs are high-quality and provides an intuitive UI for gathering feedback from human stakeholders. This helps organisations deploy production-quality generative AI solutions.

Mosaic AI Model Training enables fine-tuning of open-source foundation models with an organisation’s private data, resulting in higher-quality results for specific use cases. These fine-tuned models are fully owned and controlled by the customer, and are faster and less expensive to serve compared to larger proprietary models.

Mosaic AI Gateway provides a unified interface to query, manage, and deploy any open-source or proprietary model, allowing customers to easily switch the large language models (LLMs) powering their applications without complex code changes. It offers usage tracking, guardrails, governance, and monitoring to ensure quality and control spending.

Several Databricks customers, including Corning, Ford Direct, and Lippert, have already benefited from these new capabilities in building their generative AI applications. By leveraging the Databricks Data Intelligence Platform and Mosaic AI, they have improved retrieval speed, response quality, accuracy, and confidence in deploying to production.

The new Mosaic AI capabilities are part of Databricks’ ongoing commitment to helping customers harness the power of generative AI while maintaining data privacy, quality, and cost-effectiveness. As organisations continue to explore AI’s potential, Databricks aims to provide the tools and platform necessary for building enterprise-ready AI applications.

The post Databricks Enhances Mosaic AI for Enterprise-Ready AI Applications   appeared first on AIM.