Snorkel AI builds training datasets and simulated work environments for AI labs and companies developing models and software agents. Its data products combine expert demonstrations, scoring rules, and quality checks, while runnable environments let agents practice tasks such as coding or using business software. The company pairs domain specialists with its own data development technology to generate examples, evaluate model behavior, and refine weak areas through repeated feedback. Customers buy ready-made datasets or commission tailored data and evaluation work, and enterprises also use its specialized agent solutions. Snorkel AI was founded in 2019 and is headquartered in San Francisco, California.
Work here?Claim your company
Industries
Company Size
201-500
Company Stage
Late Stage VC
Total Funding
$586.2M
Headquarters
San Francisco, CA
Founded
2019
See people who can refer or advise you
Help us improve and share your feedback! Did you find this helpful?
Total Funding
$586.2M
Above
Industry Average
Funded Over
6 Rounds
Snorkel AI boosts multiagent evaluation accuracy and efficiency. With the integration of Prisma AIRS AI Gateway, Snorkel AI transforms its debugging workflow, gaining unprecedented visibility into multi-agent evaluations. This leap forward not only streamlines operations but also enhances the quality of outputs. LearnAI Tools Key facts. * Snorkel AI achieved a 20% accuracy boost, enhancing competitive edge in agent evaluations. * AI Gateway enables 2x faster issue detection, reducing time-to-market for new features significantly. * Visual execution paths reveal hidden vulnerabilities, improving debugging efficiency and reliability. * Faster iteration cycles cut development costs, allowing for more resources to be allocated to innovation. * Enhanced visibility into agent performance indicates a strategic shift towards data-driven decision-making. Summary. Snorkel AI faced challenges in debugging their multi-agent evaluation workflows, which involved processing thousands of results weekly. They implemented Prisma AIRS AI Gateway to enhance visibility into agent execution paths, resulting in a 20% accuracy boost in evaluations and a significant reduction in issue detection time. Background. Snorkel AI operates in the artificial intelligence sector, focusing on evaluation workflows that review expert-generated labels for specialized domains. Prior to deploying AI Gateway, the team struggled with a fragmented debugging process, relying on scattered logs and manual reconstruction of agent execution paths. Challenge. The primary challenge was the inefficiency in identifying and resolving issues within their multi-agent evaluation system. The existing debugging workflow was cumbersome, making it difficult to trace errors and understand the decision-making process of agents. LearnAI Tools Solution. The team implemented Prisma AIRS AI Gateway, which provided a visual hierarchy of agent calls, subagent interactions, and tool invocations displayed as a nested tree. This allowed for individual prompt-response inspection at any node and enabled custom tagging and filtering to identify patterns across traces. Side-by-side trace comparisons facilitated understanding of divergences between successful and failing executions. Results. The deployment of AI Gateway led to measurable improvements in Snorkel AI's evaluation workflows: * A 20% increase in accuracy for agent evaluations, enabling the team to identify and address previously hidden edge cases. * A twofold increase in the speed of issue detection during development, with problems surfacing immediately in traces instead of requiring extensive log analysis. * A significant reduction in iteration cycles, with tasks that previously took hours of manual work now completed in minutes through visual inspection. Key insights. Implementing a visual debugging tool can drastically improve the efficiency of identifying and resolving issues in complex systems. Enhanced visibility into execution paths allows teams to catch errors earlier and iterate more quickly. Customer testimonial. No direct quote is available from the source material. Discover more Find Study Aids Hire Lawyers Start an LLC Prisma AIRS AI Gateway Technologies LearnEngineering Reinforcement Learning from Human Feedback (RLHF) Key concepts. Multi-Agent Question-Answer Validator debugging workflows execution paths nested tree visualization custom tagging and filtering trace comparison accuracy boost issue detection Definitions. * Multi-Agent Question-Answer Validator - A tool used by Snorkel AI to process and validate thousands of results weekly for expert-generated labels. * Reinforcement Learning from Human Feedback (RLHF) - A machine learning approach that uses feedback from human evaluations to improve the performance of AI models. * AI Gateway - A feature of Prisma AIRS that provides visual hierarchy and insights into agent execution paths. * debugging workflow - The process of identifying and resolving issues within software systems, often involving the analysis of logs and execution paths. * aggregate metrics - Summary statistics that provide an overview of performance but may obscure individual case details. Use cases. * | Improving accuracy in agent evaluations * | Faster issue detection during development * | Visual inspection of execution paths * | Identifying patterns across traces * | Comparing successful and failing execution paths Frequently asked questions. What is the main benefit of using Prisma AIRS AI Gateway? The main benefit is the enhanced visibility it provides into agent execution, allowing teams to quickly identify and resolve issues that were previously hidden in logs. How does AI Gateway improve debugging workflows? AI Gateway improves debugging by visualizing execution paths as nested trees, enabling easier inspection of individual calls and comparisons between successful and failing traces. What kind of improvements did Snorkel AI see after implementing AI Gateway? Snorkel AI experienced a 20% boost in accuracy for agent evaluations and doubled the speed of issue detection, significantly reducing the time needed for manual log analysis. Can AI Gateway help in identifying edge cases? Yes, AI Gateway allows teams to catch and fix edge cases that were previously invisible in aggregate metrics, leading to more robust evaluation workflows. What is the significance of visualizing execution paths? Visualizing execution paths helps teams understand the flow of operations within agents, making it easier to pinpoint where things go wrong and facilitating faster troubleshooting.
SREGym at Frontier Data Summit: community, conversations, and what's next. Conversations about AI for SRE, its growing community, and a look ahead to SREGym 2.0. SREGym had a wonderful time at the Frontier Data Summit, a one-day, invite-only event organized by Snorkel AI. The summit brought together researchers and practitioners working on benchmark research that shapes the AI frontier. SREGym were delighted to participate and present SREGym at the poster session. The summit was very well organized, featuring an inspiring keynote, "The Research Era of AI Data," by Alex Ratner, Co-founder and CEO of Snorkel AI, along with a series of engaging fireside chats. One highlight was "How Benchmarks Are and Aren't Used in Model Development Today," featuring Braden Hancock, a mentor to the SREGym team, and David Hall. SREGym thoroughly enjoyed the talks and the thoughtful discussions. Interestingly, this was Tianyin's first-ever AI event. Coming from a systems research background, he found the mindsets and perspectives of the AI community both refreshing and surprisingly different. He was also struck by the breadth and diversity of the community. His personal highlight was the opportunity to reconnect with and learn from several UIUC students and alumni attending the summit. It was wonderful to see familiar connections across such a vibrant research community. SREGym is very well received at the summit. Throughout the poster session, many attendees stopped by to learn about SREGym and discuss the challenges and opportunities of AI for SRE and its evaluation. The conversations ranged from AI researchers curious about the SRE domain to core AI SRE researchers such as colleagues from ORCA-bench and Datadog. Jackson did an outstanding job presenting SREGym (according to Tianyin) and engaging with attendees from a wide range of backgrounds. Much of his effort went beyond explaining the project itself: he introduced AI researchers to the SRE domain, helped them understand its unique challenges, and made the case for why SRE is an important frontier for AI research and evaluation. These conversations sparked thoughtful discussions about the broader implications of its work and opened up opportunities for future collaborations. Some of the most memorable moments came from meeting both current and potential users of SREGym. Colleagues from Datadog made its day by sharing that they had been using SREGym for Datadog's internal agents. Another fun moment came when a colleague tried to recruit Jackson during the event. Tianyin couldn't help but laugh and feel proud. One of SREGym's strengths has always been the community SREGym is building together. SREGym now has more than 60 contributors, who have helped expand the benchmark with new problems, develop features, and fix bugs. Its Slack community has grown to 280 members, bringing together researchers, engineers, and practitioners who share an interest in advancing AI SRE. What excites SREGym even more is seeing collaborations grow beyond SREGym itself. Its community has sparked several new projects and research initiatives, some developing independently of the SREGym team. This is what makes an open research community truly valuable: not just the work SREGym accomplish together, but also the new ideas, collaborations, and opportunities that grow from it. The most frequently asked question at its poster was: "When is SREGym 2.0 coming?" SREGym were particularly excited by Alex Ratner's vision of Data 2.0, presented in his keynote, which resonated strongly with its own aspirations for SREGym. One of SREGym's core missions is to help push the boundaries of frontier AI through increasingly challenging and meaningful evaluations. Keeping pace with rapidly advancing AI capabilities is not without challenges. As models become more capable, its benchmarks must evolve to continue challenging them and revealing where meaningful progress is needed. But that's exactly what excites SREGym - there are always new mountains to climb. Its team has been making great progress on SREGym 2.0 through a new technique SREGym is calling environment scaling, and SREGym hope to release its 2.0 soon. Stay tuned! Jackson made a great poster: SREGym thank Snorkel AI again for organizing this great summit and inviting SREGym to be a part of it.
Defense, data, and deep tech draw major capital this week. Five raises totaling over $1.5B this week show investors concentrating on AI infrastructure, autonomous defense systems, and science-driven platforms with verifiable traction. Defense AI commands a serious valuation. TEKEVER's $580 million Series D, reported by Tech.eu, values the Portuguese-founded autonomous drone and AI systems company at $6.4 billion [[1]]. The round followed the company's selection for the British Army's £400 million CORVUS programme - a government contract that gave investors a hard signal of real-world demand before they committed capital at that scale. The lesson for founders is structural: sovereign defense contracts function as a form of non-dilutive revenue validation. Investors in hardware-intensive, regulated categories increasingly want to see a government or enterprise anchor before writing a large check. If your company operates near defense, critical infrastructure, or aerospace, a contract win - even a pilot - may now be a prerequisite for a growth-stage raise rather than a bonus. Rapid ARR growth compressed the seed-to-A timeline. Chamelio returned to market just five months after closing its seed round, raising a $26 million Series A led by Entrée Capital, as Tech Startups reported [[2]]. The company attributed the accelerated timeline to ARR quadrupling over those five months. Bright Pixel Capital joined as a new investor alongside existing backers Work-Bench and Emerge Ventures. Five months between rounds is unusually short. What made it possible was not a product pivot or a new market narrative - it was a specific, auditable revenue metric moving in one direction. For founders at the pre-seed or seed stage, this round underscores that vertical AI agents with demonstrated enterprise adoption can compress normal fundraising timelines significantly. The implication: do not assume a standard 18-to-24-month runway before your next raise is necessary if your numbers justify going earlier. AI infrastructure attracted broad syndicate backing. Snorkel AI raised $350 million at a $3.5 billion valuation in a round co-led by Insight Partners and S32, with participation from Greylock, March Capital, Blumberg Capital, Allegis Capital, Frontline, Standard, Third Point Ventures, and Addition, as the company announced via PR Newswire [[3]]. The breadth of the syndicate - spanning early-stage specialists, crossover funds, and late-stage growth investors - is notable. Snorkel positions itself as a data infrastructure layer for frontier AI model development. Rounds of this composition signal that investors across stages see AI data pipelines as a foundational bet, not a niche one. For founders building in AI tooling or infrastructure, the size of this syndicate suggests that the conversation with growth-stage funds can start earlier than it once did - provided the infrastructure use case is clearly tied to model development or deployment at scale. Deep science rounds are getting larger and faster. Two science-driven rounds closed in quick succession. Precision Neuroscience raised $250 million in a Series D co-led by Bill Ackman's Pershing Square at a valuation just above $1 billion, bringing its total raised to $430 million, according to The Next Web citing a New York Times report [[4]]. Enveda, which uses AI to identify drug candidates in natural compounds, raised $311 million in a Series E at a $2 billion valuation led by Catalio Capital Management, as TechCrunch reported - a figure that doubles Enveda's valuation from twelve months prior [[5]]. Both rounds involve categories - brain-computer interfaces and AI-accelerated drug discovery - that carry long development timelines and high regulatory risk. What is drawing institutional capital in despite those risks appears to be the combination of AI-enabled speed (compressing discovery cycles) and the involvement of credible scientific and financial co-leads who can hold positions across multi-year horizons. For founders in biotech, neurotech, or other deep science verticals, the takeaway is that the investor pool has expanded: crossover funds and high-profile names outside traditional life sciences are now writing significant checks. This week, if you are raising. * If you are raising in defense, autonomous systems, or regulated hardware, pursue a government or enterprise anchor contract before approaching growth-stage investors - it is increasingly a prerequisite, not a differentiator. * If your ARR has multiplied in the past quarter, run the numbers on whether your traction already justifies returning to market ahead of schedule rather than waiting out a standard runway. * If you are building AI tooling or deep science platforms, map your round's potential syndicate across stage types - this week's raises show that early-stage specialists and large crossover funds are now co-investing in the same infrastructure and science bets. Sources. defense tech ai infrastructure biotech series a
AI agents and infrastructure dominate a week of nine-figure rounds. Five deals totaling over $5B closed in a single week, revealing a market that rewards vertical AI agents with real revenue and infrastructure bets at almost any scale. The stack gets funded from bottom to top. Three of this week's largest rounds sit at different layers of the AI supply chain, and together they sketch a picture of where institutional capital is concentrating. Crusoe, an AI infrastructure provider, announced the initial close of an anticipated $3.9 billion Series F at a $30.9 billion valuation, with Nvidia joining as a backer, as Channel Insider reported [[3]]. Snorkel AI, which builds data pipelines for training frontier models, closed a $350 million round co-led by Insight Partners and S32 at a $3.5 billion valuation, with a broad syndicate including Greylock and Third Point Ventures, according to a company press release [[4]]. Then Factory, which builds AI coding agents that sit on top of that infrastructure and data, raised $200 million from a group that includes Blackstone and Salesforce CEO Marc Benioff as an angel, as The Terminal reported [[1]]. Read together, these deals are not coincidental. Compute, data, and application-layer tooling are all being capitalized aggressively and simultaneously. For a founder raising at pre-seed or seed in the AI space, the implication is structural: investors are building conviction across the entire stack, not just at the model layer. A startup that connects clearly to one of those layers - faster training data, cheaper inference, or automated developer workflows - is speaking a language that is actively being funded. Factory's valuation jump resets expectations for agent Startups. Factory's trajectory this year is the kind of data point that changes how investors calibrate their own models. According to The Terminal, the company was valued at $1.5 billion in April 2026 and has now reached $5 billion - a more than triple increase in five months - on its third large check in under a year, with total funding now past $400 million [[1]]. The speed of that re-rating is notable: it happened without an exit or a public market event, purely on the basis of investor competition for a position. For founders building in the AI agent space, this sets a new benchmark for what "strong traction" can unlock in terms of valuation step-ups between rounds. It also signals that category leadership - being the recognizable name in a specific agentic workflow - may matter more right now than margin profile. What remains unclear from the available reporting is Factory's revenue trajectory or customer count, so whether this valuation reflects fundamental performance or competitive investor pressure is an open question. Vertical AI agents are closing rounds fast. Chamelio, a legal AI startup, raised a $26 million Series A led by Entrée Capital just five months after its seed round, according to Tech Startups [[2]]. The company says its annual recurring revenue quadrupled over that same period. Work-Bench, Emerge Ventures, and Bright Pixel Capital also participated. The new capital is earmarked for expanding the product and growing the team, though specific deployment plans were not detailed in the reporting. Five months from seed to Series A is a compressed timeline, and the ARR growth figure is the clearest explanation for why investors moved that quickly. For founders in vertical AI - legal, finance, HR, compliance - Chamelio's round reinforces a pattern that has appeared across several sectors this year: when a narrow AI agent demonstrably reduces cost or time in a high-stakes professional workflow, revenue can scale fast enough to justify an accelerated round. The sector specificity is a feature, not a constraint, because it makes the value proposition measurable. European defense tech draws sovereign and institutional capital. Tekever, a Portuguese-British autonomous drone startup, closed the first tranche of a $580 million Series D at a $6.4 billion valuation, with UC Investments and Baillie Gifford leading, as Tech Startups reported [[5]]. The raise comes as European governments accelerate spending on homegrown military technology, drones, and AI-enabled defense systems. This round is somewhat separate from the agent and infrastructure story, but it points to a macro dynamic worth tracking: government procurement tailwinds can compress the sales cycle and de-risk revenue projections in ways that make institutional investors more comfortable writing large checks. For founders in autonomous systems, robotics, or any dual-use AI category with a clear government customer, European defense budgets are now functioning as a credible demand signal that investors can underwrite. This week, if you are raising. * If your startup sits at any layer of the AI stack - compute, data, or application - map that position explicitly in your pitch, because investors are building coordinated conviction across all three layers simultaneously. * Chamelio's five-month seed-to-Series A path shows that measurable ARR growth in a vertical AI workflow can compress your timeline to the next round; track and lead with that metric in every investor conversation. * Factory's valuation tripling between rounds on investor competition alone means category positioning matters acutely right now - if you are not actively working to be the named leader in your specific agentic niche, a competitor is. Sources. ai agents startup funding infrastructure defense tech
Daily research news online. The global MR industry's daily paper since 2000. Follow DRNO on... Funding values AI data lab Snorkel at $3.5 billion. September 24 2026 Redwood City, CA-based Snorkel AI, whose data platform and agentic, research-driven data factory help inform AI model development, has raised $350 million in Series C funding, at a $3.5bn valuation. The firm says it will recruit researchers and engineers, to meet booming demand for complex training data. Snorkel is led by CEO Alex Ratner, and has developed an agentic data factory and expert Data-as-a-Service platform delivering data, benchmarks, evaluations and custom environments for frontier and agentic AI model development. Clients are AI labs and global enterprises in sectors including healthcare, law and software engineering, as well as the US Federal Government. Snorkel, which says it expects to reach profitability this year, combines human experts and thousands of specialized AI models and agents to create and check datasets. Quoted by Reuters, Ratner said experts devise design scenarios, tasks and grading rubrics, while AI automates much of the labor-intensive quality assurance process. The firm has built a network of 'tens of thousands of specialists' in coding, law, medicine and other fields. The round was co-led by Insight Partners and S32, with participation from Addition, March Capital, Blumberg Capital, Allegis Capital, Frontline, Standard, Third Point Ventures, Greylock, Lightspeed, GV, Factory, Prosperity7, Walden Catalyst and Wells Fargo. Home page: www.snorkel.ai. All articles 2006-23 written and edited by Mel Crowther and/or Nick Thomas, 2024- by Nick Thomas, unless otherwise stated. Most viewed items in the last week... Each (*) indicates > 1,000 views. Select a region below...
Find jobs on Simplify and start your career today
Don't see your dream role? Check out thousands of other roles on Simplify.Browse all jobs
Industries
Company Size
201-500
Company Stage
Late Stage VC
Total Funding
$586.2M
Headquarters
San Francisco, CA
Founded
2019
Find jobs on Simplify and start your career today
Don't see your dream role? Check out thousands of other roles on Simplify.Browse all jobs