- Snorkel AI introduces Snorkel Custom, a comprehensive platform for enterprises to optimize Large Language Models (LLMs) for custom AI use cases.
- The offering combines Snorkel Flow, an AI data development platform, with expert support to programmatically develop and fine-tune data for LLMs.
- Enterprises can adapt any LLM for their unique needs, with native integrations for seamless model integration.
- Snorkel Custom streamlines AI development with collaborative evaluation workshops, guided data and LLM development, and model cost optimization.
- A case study with Wayfair demonstrates Snorkel Custom’s success, enabling rapid deployment of AI solutions and enhanced business agility.
Main AI News:
In a bold move to revolutionize the landscape of enterprise AI development, Snorkel AI has unveiled Snorkel Custom, a groundbreaking service and platform designed to empower enterprises in leveraging their data to optimize Large Language Models (LLMs) and expedite the delivery of high-quality AI solutions. Recognizing the inherent challenge that LLMs often do not meet the specific needs of individual enterprises right out of the box, Snorkel Custom offers a comprehensive solution. By amalgamating the cutting-edge AI data development platform, Snorkel Flow, with hands-on guidance from Snorkel’s team of machine learning experts, this innovative offering enables enterprises to systematically develop and fine-tune data, thereby ensuring the seamless alignment of LLMs with their unique use cases.
“In today’s competitive landscape, the efficacy of AI solutions hinges entirely on the quality of the underlying data utilized to calibrate and refine the models,” remarked Alex Ratner, CEO and co-founder of Snorkel AI. “To bridge the gap between captivating AI demonstrations and tangible ROI in production environments, enterprises must master the art of curating, labeling, and refining data tailored to their specific requirements and environments.”
Snorkel Custom empowers enterprises to customize any LLM to suit their distinct use cases. Moreover, in a bid to streamline accessibility for customers, Snorkel is committed to providing an extensive library of native integrations with leading LLMs. As part of this endeavor, Snorkel has announced the integration of Google Cloud’s Gemini models into its platform, further enhancing the array of options available to enterprises.
Driving Forward the Era of Production-Ready Generative AI
The burgeoning interest in Generative AI has spurred enterprises to expedite the deployment of AI applications primed for production environments. However, achieving optimal performance on bespoke use cases necessitates fine-tuning LLMs using an enterprise’s proprietary data. Traditionally, this data development process has been laborious, costly, error-prone, and potentially non-compliant with data regulations. Enter Snorkel Flow—a modern, automated solution that streamlines key data operations such as labeling, sampling, filtering, and slicing, akin to software development practices.
Paired with a structured engagement framework, Snorkel Custom offers a holistic approach to AI development, encompassing:
- Collaborative evaluation and benchmark workshops tailored to create bespoke benchmarks for each use case, leveraging Snorkel’s expertise in LLM evaluation and data operations alongside customers’ domain knowledge.
- Guided data and LLM development led by Snorkel’s experts, ensuring end-to-end delivery of finely-tuned LLMs aligned with production-level benchmarks.
- Model cost optimization and serving options, including the distillation of LLMs into specialized “small language models” (SLMs) to enhance task-specific accuracy while optimizing costs.
- Implementation of Snorkel Flow as the cornerstone for programmatic AI data development, facilitating ongoing adaptation, maintenance, and auditability of developed LLMs, with optional support for teams transitioning to self-service models.
Unleashing the Power of Snorkel Custom: A Case Study with Wayfair
Illustrating the transformative impact of Snorkel Custom, Wayfair—a leading e-commerce giant—has witnessed unparalleled success in deploying AI solutions at scale. With over 10,000 product tags spanning a vast catalog of more than 30 million products, Wayfair faced the daunting challenge of ensuring product relevance in customer searches. Traditionally, their tagging process relied heavily on manual annotation, a cumbersome and time-intensive endeavor.
However, by embracing Snorkel Flow, Wayfair revolutionized its approach to data development. Through a collaborative effort with Snorkel’s team, Wayfair implemented a programmatic data development strategy tailored to its unique requirements. The result? A tenfold increase in the speed of AI service deployment enables rapid adaptation to evolving market trends.
Margaret Pierson, Director of Machine Learning at Wayfair, lauded Snorkel’s partnership, emphasizing its pivotal role in driving business agility and unlocking tangible value. “With Snorkel, we’ve transcended traditional constraints, delivering superior AI solutions in days rather than months,” remarked Pierson. “Moreover, the seamless integration of Snorkel Flow into our model lifecycle ensures that we remain at the forefront of innovation, empowering us to swiftly capitalize on emerging trends.”
Conclusion:
Snorkel Custom’s introduction marks a significant advancement in enterprise AI development, offering organizations a streamlined approach to harnessing the power of AI for bespoke use cases. By providing a comprehensive solution that combines cutting-edge technology with expert guidance, Snorkel AI is poised to drive transformative change in the market, empowering enterprises to unlock new levels of efficiency, agility, and innovation.
