New Snorkel Flow Release Empowers Enterprises to Harness Their Data for Custom AI Solutions

News provided by

Apr 24, 2024, 09:00 ET

Snorkel AI's new release streamlines secure, programmatic data development so enterprises can cross the chasm to production-quality AI

REDWOOD CITY, Calif., April 24, 2024 /PRNewswire/ -- Snorkel AI today announced the latest release of its flagship AI data development platform, Snorkel Flow, to further accelerate large language model (LLM) customization with enterprise data. Snorkel Flow's latest release advances its programmatic approach to AI data development by introducing critical enterprise extensibility, readiness and collaboration features as well as broader support for multimodal data.

While early adopters have realized some initial value from off-the-shelf LLMs, the more exciting potential of enterprise AI lies in the use of custom LLMs to perform specialized, business-specific tasks. To create custom LLMs, data from across the enterprise, often unstructured and highly sensitive, must be carefully selected, filtered, labeled and curated – and properly applied to a flexible range of base LLMs. Snorkel Flow is a scalable platform which makes this AI data development programmatic, just like software development, enabling industry leaders like Wayfair, BNY Mellon and Chubb to successfully build customized, production AI by operationalizing their data.

"Enterprises are quickly hitting a wall with what they can achieve using off-the-shelf LLMs, and are seeing that the next wave of value will be unlocked by tuning LLMs on their unique data and use cases," said Alex Ratner, co-founder and CEO, Snorkel AI. "As base LLMs become pervasive, including powerful open source options like Llama 3, the speed and accuracy with which data is continuously labeled and curated for fine-tuning and aligning LLMs becomes the key differentiator."

With features for enhanced connectivity, enterprise readiness, and human-in-the-loop collaboration, this release establishes Snorkel Flow as the central enterprise AI data development platform.

Flexible data and LLM connectivity enhance Snorkel Flow's position as a central and open AI data development platform that can be used to fine-tune any off the shelf LLM with all data types, from any enterprise source:

New templated LLM integrations for Google's Gemini model family and Meta's Llama 3 add to an existing library of native LLM integrations.
Improved SDK functionality enables custom integration with any LLM.
New data source integrations with Databricks Unity Catalog, Vertex AI and Microsoft Azure Machine Learning increase the ease with which enterprises can label, curate, and develop data for tuning LLMs.
New features for retrieval augmented generation (RAG) include the fine-tuning of embeddings and retrieval models.
Multimodal (image/computer vision) use case support for programmatic data labeling to meet the wave of interest in other data modalities, such as image, video, and audio.

Enterprise Readiness: As many GenAI startups lack the core enterprise readiness features required to operate in production, Snorkel's new release doubles down on critical enterprise readiness functionality:

New role-based access controls (RBAC) solves one of AI's biggest challenges: protecting PII and other sensitive data. RBAC granularity gives admins more control over who has access to what data, and who can upload data for AI development.
On-premise and air-gapped Foundation Model access ensures compliance by providing secure access to foundation models directly from Snorkel Flow.
Expanded end model deployment support unlocks model deployment across a variety of enterprise environments, including new operating systems (e.g. RHEL, CentOS) and hardware configurations (e.g. GPU).

Subject matter expert evaluation, annotation, and collaboration: Connections between subject matter experts (SMEs) who understand the data and can evaluate model outputs is critical to achieving production quality AI. The new release of Snorkel Flow extends in platform support for SME annotation, evaluation, and collaboration.

Reviewer workflow management gives annotation reviewers more efficient triage and easier correction of labeling errors
More performant data exploration allows SMEs to explore their data with powerful filters/viewers and instantaneous feedback.

Snorkel Flow and Snorkel Custom are both generally available. Pricing is based on the use case; to learn more, contact [email protected].

Resources:

About Snorkel AI
Snorkel AI has been pioneering programmatic AI data development since 2015, when its founders launched the Snorkel research project out of the Stanford AI Lab. Working in partnership with teams from Apple, Google, and other industry and government sponsors, the team quickly delivered results that accelerated AI application development by 10-100x or more. This transformative approach served as the design point for Snorkel Flow, an enterprise AI data development platform. The team has published more than 100 peer reviewed research papers and is used in production by Fortune 500 companies including BNY Mellon, Wayfair, Chubb, as well as across the federal government. Snorkel AI operates out of San Francisco, New York and London.

Media Contact: [email protected]

SOURCE Snorkel AI