SAN FRANCISCO & REDMOND, Wash.: Scale AI, the data engine that powers the most advanced AI applications, today announced a collaboration with Microsoft at Microsoft Ignite to deliver Scale’s Enterprise Generative AI Platform (EGP) on Microsoft Azure, giving customers direct access to tools and services to customize large language models (LLMs) and build and deploy robust Generative AI applications.

With this integration, enterprises can scale more use cases to production faster, achieve state-of-the-art model performance on their use cases, and maximize ROI on their Generative AI investments. Scale’s EGP – which helps customers implement retrieval augmented generation (RAG), fine-tune LLMs, perform expert Test and Evaluation (T&E), and build and deploy Generative AI applications – will be available directly within Microsoft Azure.

As enterprises look to adopt Generative AI, they are finding that even the most advanced foundation models are not capable of high-quality performance for their specific use cases, and many enterprises lack the tools needed to customize and scale models to production. While sophisticated Generative AI models like OpenAI’s GPT-4 are widely available today, only 10% of enterprises actually have AI models in production. This is why Scale has collaborated with Microsoft to help enterprises more easily build and deploy customized LLM applications.

“At Scale, our mission is to enable companies in every industry to harness the power of AI,” said Arun Murthy, Chief Product and Technology Officer, Scale AI. “Since Scale’s inception, we’ve been doing this for companies like OpenAI, Meta, Accenture, Brex, and Instacart. Now, with our Microsoft collaboration, businesses get access to battle-tested tools and expertise to harness their proprietary data to build Generative AI applications that help them maximize the ROI of their Generative AI investments.”

“With this next generation of AI, we have an extraordinary opportunity to help organizations build customized Generative AI applications to transform every part of their business,” said Zia Mansoor, Corporate Vice President, Data and AI, Microsoft. “Our integration with Scale AI will help Microsoft Azure customers unlock a new wave of AI-powered innovation.”

Generative AI Made for Customers

The new platform delivers a fully integrated experience for fine-tuning, managing, and building LLM applications within Azure. Customers will be able to purchase Scale’s EGP through Azure Marketplace, leveraging their existing Azure agreements.

Microsoft Azure customers will benefit from:

• Management of model fine-tuning, versioning, endpoints, performance evaluations, and training data in a single platform.
• Integration with the Scale Data Engine, with access to subject matter experts and the tools to bring your own experts to train, align, and evaluate models.
• Ability to customize commercial and open-source models, including Azure-hosted OpenAI models through Azure OpenAI Service.
• Advanced tools for RAG, including APIs and workflows for chunking and indexing structured and unstructured data and storing embeddings in vector stores.
• Scale’s expert red teaming services to identify risks and vulnerabilities and provide actionable recommendations on how to mitigate them.

For more information on how to get started with Scale on Azure, visit https://scale.com/partners/microsoft.

About Scale AI

Scale AI unlocks AI for every industry, transforming how organizations build and deploy Generative AI. Its data engine is powering the most advanced generative and computer vision models with high-quality data to provide the blueprint for organizations to apply AI to their unique use cases. Scale is trusted by the most ambitious companies across industries, including Meta, Microsoft, U.S. Army, DoD’s Defense Innovation Unit, Open AI, Cohere, Anthropic, General Motors, Toyota Research Institute, Brex, Instacart and Flexport. Scale was founded in 2016 and is valued at $7.3 billion, backed by Founders Fund, Accel, Index, and Tiger Global.

Media Contact
Anna Franko
press@scale.com

Fonte: Business Wire