Snorkel AI announced Application Studio, a visual builder with templated
solutions for common AI use cases based on best practices from hundreds of
deployments and research at top academic institutions over the last six
years. Application Studio is in preview and will be generally available
later this year within Snorkel Flow, the first AI development platform to
programmatically label data and train, deploy and analyze models
iteratively.
"Over the years we've heard a clear refrain from enterprises working to deploy
AI: data is the blocker. In many settings today-for example, ones where
privacy, expertise or speed are essential-even the largest organizations
can't afford to manually label the volume of data that modern machine learning
approaches require," said Alex Ratner, co-founder and CEO of Snorkel AI.
"Snorkel Flow's programmatic approach to training data labeling and model
development uniquely unlocks these use cases and a whole new way
to rapidly and iteratively develop AI applications-which we're now excited
to make increasingly templatized and fast to deploy with Application
Studio."
According to Cognilytica's "Data Preparation & Labeling for AI 2020 Report,"
80 percent of AI development time is spent on gathering, organizing and
labeling data manually which is used to train machine learning models.
Hand-labeling is notoriously expensive and slow with limited ability for
development teams to build, iterate, adapt or audit applications in a
systematic and privacy-compliant manner. The training data bottleneck has
made AI application development an impractical endeavor, and 87 percent of the
data science projects never make it into production.
"Snorkel AI addresses key points of pain for enterprises that need to
digitally transform their businesses with production ML. Their data teams
struggle to build, train and deploy accurate models at scale because
the coding is complex and data volumes keep rising. They need to optimize
their use of existing code, accelerate model development and organize
training data more efficiently. They also need to collaborate on a common
platform that supports the full ML lifecycle," said Kevin Petrie, Vice
President of Research at Eckerson Group.
Snorkel Flow makes it possible for data scientists, developers and subject
matter experts to rapidly create and manage training data, train custom
machine learning models, analyze and iterate to systematically improve and
adapt and deploy AI applications quickly. With Snorkel Flow, organizations have
achieved state-of-the-art machine learning model accuracy in days and
10-100x reductions in development time. Customers include two of the
three top US banks, global insurance, biotech and telecommunications providers
and several government agencies.
With the introduction of Application Studio, Snorkel AI lets enterprises
develop AI applications faster than ever before. Application Studio
provides data scientists, developers and subject matter experts with:
- Pre-built solution
templates: Pre-built solution templates based on industry-specific use
cases such as contract intelligence, news analytics and customer
interaction routing or common AI tasks such as text and document
classification, named entity recognition and information extraction, give
enterprise data science teams a head start in developing their own
applications. Packaged application-specific pre-processors, programmatic
labeling templates, models and features make customizing applications
as easy as dragging and dropping new logic into the application flow.
- High-performance models:
Enterprises can use their own private data, labeled programmatically,
to train state-of-the-art, open source model libraries available in the platform.
Programmatic labeling replaces weeks or months of costly hand-labeling
yielding highly accurate model performance.
- Collaborative
workflows: Application Studio allows for an intuitive decomposition of
complex applications into modular parts so that data scientists,
developers and domain experts can collaborate easily and efficiently.
- Auditable and adaptable
capabilities: With Application Studio, the entire pipeline from training
datasets to user contributions is versioned and can easily be audited.
With a few lines of code, the applications can adapt to new data or goals.
Unlike other application platforms that rely on hand-labeled data,
there is no need to start from scratch.
- Data privacy at enterprise
scale: Data breach and bias risks are the largest blockers to
applying machine learning to many problem domains and sectors. With
Application Studio, training data labeling and management are not only
kept in-house but also can be done without humans needing to view the
majority of the data-setting a new high bar for practical, private machine
learning.
Snorkel AI Raises a $35 Million Series B Growth
Funding Round
Today Snorkel AI also announced $35 million in Series B funding, bringing the
total raised to $50 million. This round was led by Lightspeed Venture
Partners; previous investors Greylock, GV, In-Q-Tel and Nepenthe Capital
and new investors Walden and funds and accounts managed by BlackRock
also participated. The company will use the funding to continue scaling
its world-class engineering team and bringing its technology to leading
enterprises.
Ravi Mhatre, Partner at Lightspeed Venture Partners and Snorkel AI Board
Member, said: "Enterprises are spending billions of dollars to put AI to
use today, and the AI technology market is expected to hit the half
a trillion dollar mark in a few years. Snorkel AI is solving one of the
biggest problems in AI - the data. With Snorkel Flow, organizations of all
sizes, including some of the world's most sophisticated ones, are
applying AI to mission-critical challenges and building solutions
previously not possible. The traction has been incredible and they're just
getting started."