--- title: What Frontier Labs Buy: A Data Deep Dive with Handshake... description: Handshake AI's training-data business grew from launch to $1 billion in gross annualized revenue in about a year. This podcast walks through live demos of the... canonical: https://daily.dev/posts/what-frontier-labs-buy-a-data-deep-dive-with-handshake-ai-garrett-lord-team-9cq9rro8c twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: What Frontier Labs Buy: A Data Deep Dive with Handshake AI — Garrett Lord & Team | daily.dev og:description: Handshake AI's training-data business grew from launch to $1 billion in gross annualized revenue in about a year. This podcast walks through live demos of the... og:url: https://daily.dev/posts/what-frontier-labs-buy-a-data-deep-dive-with-handshake-ai-garrett-lord-team-9cq9rro8c og:image: https://api.daily.dev/og/posts/9CQ9rrO8c.png og:image:alt: What Frontier Labs Buy: A Data Deep Dive with Handshake AI — Garrett Lord & Team og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # What Frontier Labs Buy: A Data Deep Dive with Handshake AI — Garrett Lord & Team **[Latent Space](https://daily.dev/sources/latentspace)** · 4 min read · 1 upvotes · 0 comments ## Summary Handshake AI's training-data business grew from launch to $1 billion in gross annualized revenue in about a year. This podcast walks through live demos of the reinforcement learning environments frontier labs buy from Handshake. In a finance demo built on the ATLAS-Finance benchmark, an agent drafts a bank workbook but mishandles a tax-treatment rule despite being able to explain it when asked directly. In a biology demo using the VIALS benchmark, a model asked to pick the two brain-tissue slides showing the most advanced vessel necrosis picks those two and adds a third slide, G. A coding segment shows the team inspecting an agent's file access and shell history to catch gaming behavior invisible in its final reply. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How much revenue does Handshake AI make from training data for frontier AI labs? Handshake's AI training data business grew from launch to $1 billion in gross annualized revenue within roughly a year, according to reporting cited alongside the company's own data deep dive. The business spans multiple verticals including finance, biology, coding, and alignment, building reinforcement learning environments and benchmarks that frontier labs purchase to train and evaluate models. _Developers tracking the business of AI training data can follow deep dives like this on daily.dev._ ### What is the VIALS benchmark for evaluating vision-language models? VIALS is Handshake AI's benchmark for testing how well models interpret scientific images, such as gels and brain-tissue slides used in biology research. In one demonstrated task, a model asked to identify the two brain-tumor tissue slides showing the most advanced vessel necrosis correctly selected both right slides but added an extra incorrect one, illustrating a gap between human expert judgment and model performance on specialized image interpretation. _Those evaluating vision-language models on domain-specific imagery can dig into benchmarks like this via daily.dev._ ### What mistakes do AI agents make when handling financial modeling tasks in a simulated investment bank? In Handshake's ATLAS-Finance benchmark, an agent tasked with building a bank deal workbook correctly carries a goodwill figure from a client email into its Python calculations and subtracts it, but fails to recognize that a specific line of code is where a tax-treatment rule should apply, even though it can explain that rule correctly when asked directly, showing a gap between stated knowledge and applied execution. _Teams deciding whether AI agents are ready for real financial modeling work can track evals like this on daily.dev._ ## Similar posts on daily.dev - [AI data labeler Handshake buys Cleanlab, an acquisition target of multiple others](https://daily.dev/posts/ai-data-labeler-handshake-buys-cleanlab-an-acquisition-target-of-multiple-others-ocbctesms) · TechCrunch · 0 upvotes · 0 comments - [The New AI Problem Is a Lack of New Data](https://daily.dev/posts/the-new-ai-problem-is-a-lack-of-new-data-4uhgaxf7l) · Kilo Blog · 2 upvotes · 0 comments --- Tags: [#machine-learning](https://daily.dev/tags/machine-learning), [#ai-agents](https://daily.dev/tags/ai-agents), [#reinforcement-learning](https://daily.dev/tags/reinforcement-learning) [View this post on daily.dev](https://daily.dev/posts/what-frontier-labs-buy-a-data-deep-dive-with-handshake-ai-garrett-lord-team-9cq9rro8c) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"What Frontier Labs Buy: A Data Deep Dive with Handshake AI — Garrett Lord \u0026 Team","url":"https://daily.dev/posts/what-frontier-labs-buy-a-data-deep-dive-with-handshake-ai-garrett-lord-team-9cq9rro8c","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/what-frontier-labs-buy-a-data-deep-dive-with-handshake-ai-garrett-lord-team-9cq9rro8c"},"datePublished":"2026-11-09T17:20:17.836Z","dateModified":"2026-11-09T17:20:50.992Z","description":"Handshake AI's training-data business grew from launch to $1 billion in gross annualized revenue in about a year. This podcast walks through live demos of the...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/2dacfa5f24b6430086aea3a50abb268f?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/2dacfa5f24b6430086aea3a50abb268f?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Latent Space","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Latent Space","logo":"https://media.daily.dev/image/upload/s--DnJ9laFj--/f_auto,q_auto/v1780213196/logos/latentspace?_a=BAMAMiWQ0","url":"https://daily.dev/sources/latentspace"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/what-frontier-labs-buy-a-data-deep-dive-with-handshake-ai-garrett-lord-team-9cq9rro8c","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"machine-learning,ai-agents,reinforcement-learning","timeRequired":"PT4M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Latent Space","item":"https://daily.dev/sources/latentspace"},{"@type":"ListItem","position":3,"name":"What Frontier Labs Buy: A Data Deep Dive with Handshake AI — Garrett Lord \u0026 Team"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/what-frontier-labs-buy-a-data-deep-dive-with-handshake-ai-garrett-lord-team-9cq9rro8c#faq","mainEntity":[{"@type":"Question","name":"How much revenue does Handshake AI make from training data for frontier AI labs?","acceptedAnswer":{"@type":"Answer","text":"Handshake's AI training data business grew from launch to $1 billion in gross annualized revenue within roughly a year, according to reporting cited alongside the company's own data deep dive. The business spans multiple verticals including finance, biology, coding, and alignment, building reinforcement learning environments and benchmarks that frontier labs purchase to train and evaluate models. Developers tracking the business of AI training data can follow deep dives like this on daily.dev."}},{"@type":"Question","name":"What is the VIALS benchmark for evaluating vision-language models?","acceptedAnswer":{"@type":"Answer","text":"VIALS is Handshake AI's benchmark for testing how well models interpret scientific images, such as gels and brain-tissue slides used in biology research. In one demonstrated task, a model asked to identify the two brain-tumor tissue slides showing the most advanced vessel necrosis correctly selected both right slides but added an extra incorrect one, illustrating a gap between human expert judgment and model performance on specialized image interpretation. Those evaluating vision-language models on domain-specific imagery can dig into benchmarks like this via daily.dev."}},{"@type":"Question","name":"What mistakes do AI agents make when handling financial modeling tasks in a simulated investment bank?","acceptedAnswer":{"@type":"Answer","text":"In Handshake's ATLAS-Finance benchmark, an agent tasked with building a bank deal workbook correctly carries a goodwill figure from a client email into its Python calculations and subtracts it, but fails to recognize that a specific line of code is where a tax-treatment rule should apply, even though it can explain that rule correctly when asked directly, showing a gap between stated knowledge and applied execution. Teams deciding whether AI agents are ready for real financial modeling work can track evals like this on daily.dev."}}]} ```