Snorkel AI

What Snorkel AI does

Snorkel AI is a privately held frontier AI data lab founded in 2019 out of Stanford AI Lab and headquartered in Palo Alto, California. The company builds expert-curated training datasets, benchmark-grade evaluation systems, and runnable RL environments for frontier AI labs and large enterprises, with a methodology rooted in programmatic data labeling, calibrated expert signal, rubric-based verification, and multi-layer quality pipelines. Its product surface centers on Snorkel Flow (the underlying AI Data Development Platform), complemented by Snorkel Evaluate for model evaluation and Snorkel Expert Data-as-a-Service, a two-sided platform that pays domain experts to contribute to AI projects. The Snorkel Data Series delivers curriculum-structured datasets (GDPval+, SWE-Bench+, CUA-Bench+, Enterprise Agentic Environments) refreshed quarterly, alongside benchmark assets such as Terminal-Bench 2.0, SlopCodeBench, Agents' Last Exam, Continual Learning Bench, and BigLaw Bench. Custom specialized-agent engagements and the open-source Harbor framework round out the offering.

The company monetizes primarily through subscription-style Data-as-a-Service and benchmark/evaluation programs, supplemented by professional services for custom agent development and enterprise AI deployments. Distribution combines direct enterprise sales (with discovery phases, pilots, and staged rollouts), AWS Marketplace and Microsoft Azure integrations, and Accenture co-sell motion for financial services. Snorkel's customer roster spans frontier AI labs (Google, Microsoft, OpenAI, Anthropic), financial institutions (BNY, an unnamed global bank), legal technology (Harvey), and large enterprises (Apple, Intel, Stanford Medicine), with 30 or more Fortune 500 attendees reported at the inaugural SnorkelCon in 2024.

Financially, Snorkel reported $148 million in ARR for 2025, up from $36.8 million in 2024, and closed a $100 million Series D in May 2025 at a $1.3 billion post-money valuation led by Addition. The company is venture-backed (Addition, BlackRock, Greylock, Lightspeed, Google Ventures, Accel, IQT, Accenture Ventures), holds SOC 2 Type 2 and HIPAA compliance, and after a 13% workforce reduction in September 2025 has re-centered around its Data-as-a-Service and specialized-agent businesses.

Name

Snorkel AI

Legal name

Snorkel AI, Inc.

Website

https://snorkel.ai

Company type

Private

Founded year

2019

Operating status

Operating

Headcount range

251–500 employees

Short description

Snorkel AI is a frontier AI data lab that builds expert-curated training datasets, rubric-based evaluation systems, and runnable RL environments for frontier AI labs and large regulated enterprises, sold via DaaS subscriptions, benchmark products, and custom agent engagements.

Ownership category

akta.pro rank

Product category

AI Data Development Platform

NAICS

Computer Systems Design and Related Services (54151), Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (518210)

SIC

Services-Computer Processing & Data Preparation (7374), Services-Computer Programming, Data Processing, Etc. (7370)

akta.pro primary industry

Domain-Specific Training Data Solutions (e.g., medical, AV, geospatial) (HDAAALAL)

Keywords

• AI data development
• Programmatic data labeling
• AI evaluation benchmarking
• Specialized AI agents
• Expert-curated datasets

Headquarters

HQ city

San Francisco

HQ country

United States

HQ region

North America

Offices2 records

Markets served

GTM type

B2B

Offering type

Software

Cost components

Personnel, Technology or R&D, Marketing or Sales, Infrastructure, Operations

Revenue model

1. Data-as-a-Service (DaaS): Expert-curated data development services delivered as ongoing programs. Includes ready-to-use Snorkel Data Series products and custom dataset development engagements. Revenue generated through project-based and recurring program contracts with enterprise clients.
2. Benchmark and Evaluation Products: Sale of benchmark-grade datasets and environments from the Snorkel Data Series catalog. Programs include Terminal-Bench+, SWE-Bench+, Enterprise Agentic Environments, CUA-Bench+, and GDPval+ products refreshed quarterly.
3. Custom Agent Development: End-to-end custom AI agent development services for specialized enterprise workflows. Covers scoping, development, deployment, and continuous iteration engagements.
4. Snorkel Evaluate: Enterprise evaluation platform product for measuring and tuning specialized AI models. New product launched in 2025 as part of expanded platform offering.

Pricing tiers

Go-to-market motion3 records

Distribution channels3 records

Marketing channels8 records

Core offering

Snorkel AI is a frontier AI data lab that develops specialized training data, evaluation systems, and runnable environments for frontier AI models and agents. Its core product is the Snorkel Flow AI Data Development Platform, supported by Snorkel Evaluate for model evaluation, Snorkel Expert Data-as-a-Service for expert-curated data programs, the Snorkel Data Series of curriculum-structured datasets and benchmarks, and Specialized Agents built for enterprise workflows.

Product overview

Snorkel AI is a frontier AI data lab offering a comprehensive data development platform and suite of specialized products. The core offering is Snorkel Flow (AI Data Development Platform), supplemented by Snorkel Evaluate for model evaluation and Snorkel Expert Data-as-a-Service for expert-curated data. The Snorkel Data Series provides curriculum-structured datasets and benchmarks including Terminal-Bench 2.0, SlopCodeBench, BigLaw Bench, SWE-Bench+, CUA-Bench+, GDPval+, and Enterprise Agentic Environments. Specialized Agents deliver custom AI solutions for enterprise workflows. All offerings are grounded in calibrated expert signal, rubric-based verification, and a data-centric development approach that enables measurable iteration and expert-trusted AI systems.

Differentiator

Problem solved

Functional benefit

Brands

• Snorkel Evaluate: Product for evaluating and tuning specialized AI models more efficiently.
• Snorkel Expert Data-as-a-Service
• Snorkel Flow
• Snorkel Develop

Products and services

• Snorkel Flow

Quantifiable outcome

• Improved legal contract extraction agent from 25% to 95% expert acceptance rate through realigned evaluation criteria and training data
• +2 more outcomes

Named customers10 records

Segments4 records

Ideal customer profiles4 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

Integration2 records

AI capability12 records

Feature5 records

Partnerships

Seven partnerships are on record, tiered core, major and minor.

• AnthropiccoreStrategic or Co-development Partner · 1 June 2026Active collaboration on Claude Code AI coding tool improvement project called 'Marlin', using feedback from approximately 1,000 contractors with software engineering backgrounds.
• HarveymajorStrategic or Co-development Partner · 11 March 2026Collaboration on BigLaw Bench research benchmark for legal AI evaluation. Snorkel AI contributed expertise in data development and evaluation methodology.
• Amazon Web Services (AWS)coreTechnology or Integration · 25 November 2024Joined AWS ISV Accelerate Program and launched Snorkel Flow availability in AWS Marketplace, enabling broader enterprise distribution through AWS customer base.
• MicrosoftcoreTechnology or Integration · 26 September 2024Strategic partnership to unlock enterprise AI potential. Snorkel AI is integrated with Azure and part of Microsoft for Startups Pegasus Program. Collaboration on enterprise AI solutions.
• BNYcoreStrategic or Co-development Partner · 9 August 2024Agreement to accelerate development of AI models for banking operations. Collaboration focused on enabling BNY to develop and deploy predictive AI models.
• Accenture VenturesmajorImplementation/ SI/ Consulting PartnerProject Spotlight accelerator program providing access to Accenture's industry expertise and enterprise client network for growth-stage software companies.
• Scale AIminorOthersMentioned as competing platform in AI training data market alongside Snorkel AI, offering up to $110 per hour for specialized AI training work.

Scale indicators8 records

Recent moves8 records

Expansion highlights7 records

Direct peers

• Labelbox: Labelbox provides a data-centric AI platform for training data creation, labeling, and evaluation, sold to ML teams at enterprises. It competes directly with Snorkel Flow as a programmatic labeling and curation platform.
• Scale AI: Scale AI is the leading AI data labeling and evaluation platform, serving frontier labs and enterprises with annotated training data and RLHF services. It directly competes with Snorkel across data labeling, evaluation, and expert-data offerings and is explicitly named as a competitor in Snorkel's market.
• Surge AI: Surge AI is a data labeling platform purpose-built for frontier AI teams, focusing on expert-labeled datasets and evaluation. It targets the same frontier-lab and enterprise customers with a similar expert-driven labeling model.
• Sama: Sama delivers AI data annotation and model validation services with a focus on high-quality, human-labeled data for ML teams. It overlaps with Snorkel's expert-curated data development and evaluation services for enterprise AI deployments.
• SuperAnnotate: SuperAnnotate offers an end-to-end data annotation and management platform for ML teams building computer vision and LLM applications. It competes with Snorkel Flow on tooling for programmatic labeling and dataset management.

Broad incumbents

• Appen: Appen is an established public data annotation vendor supplying labeled datasets for AI/ML at global scale. While less specialized in frontier model evaluation, it competes for enterprise AI training data budgets and large-scale labeling contracts.
• TELUS International (Lionbridge AI): TELUS International AI (formerly Lionbridge AI) provides large-scale multilingual data annotation and AI data services to enterprise customers. It overlaps with Snorkel's enterprise DaaS offerings but with a broader, less frontier-model-focused portfolio.
• Amazon SageMaker Ground Truth: SageMaker Ground Truth is AWS's managed data labeling service integrated into the AWS ML stack. It offers a broad, commoditized alternative to Snorkel Flow for AWS-native customers and competes for the same enterprise training-data workloads.

Emerging players

• Datasaur: Datasaur is an NLP-focused data labeling platform with LLM-assisted labeling workflows. It is an emerging alternative to Snorkel Flow for text-heavy enterprise labeling use cases.
• Weights & Biases: Weights & Biases provides experiment tracking, model evaluation, and dataset versioning tooling for ML teams. Its evaluation and dataset products partially overlap with Snorkel Evaluate and Snorkel's benchmark suite.

Market position

Strengths4 records

Weaknesses4 records

Competitive moat6 records

Key risks6 records

Key highlights7 records

Customer concentration

• Xx.com/SnorkelAI
• YouTubeyoutube.com/channel/UC6MQ2p8gZFYdTLEV8cysE6Q
• LinkedInlinkedin.com/company/home

Compliance2 records

Revenue estimate

Valuation estimate

Number of profiles

Profiles12 records

Funding overview

Funding rounds9 records

Investors27 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Snorkel AI