On Thursday, Ellis AI publicly unveiled its operations, securing $10 million in seed funding. The investment is backed by notable firms such as First Round Capital, 645 Ventures, Harlem Capital, Khosla Ventures, Thrive Capital, Slow Capital, Kearny Jackson, and Ariel Alternatives CEO Mellody Hobson.

Designed to streamline the disjointed workflow typical among private credit managers, Ellis integrates AI agents to manage various tasks, from handling documents and spreadsheets to correspondence. The company was established by Ryan Williams, recognized for co-founding the real estate investment platform Cadre with Josh and Jared Kushner in 2014. Under Williams' leadership, Cadre raised over $160 million and reached a peak valuation of $800 million before being acquired by the alternative investment firm Yieldstreet in 2024 for an undisclosed amount.

Williams pointed out the challenge he observed while at Cadre: “While the front end of private markets has become more user-friendly and accessible, the underlying operational infrastructure remains fragmented.” 

Last year, he commenced the development of Ellis with the aim of unifying the disparate software, accounting, and documentation utilized by private credit firms into a single, user-friendly platform. This innovative system not only identifies data inconsistencies but also employs AI agents to assist with portfolio monitoring and report preparation.

For instance, Williams highlighted that these AI agents can facilitate the monthly closure of a fund’s financial records. “A team typically needs to download files from multiple systems, reformat the data, verify balances, investigate discrepancies, and manually re-enter information. In many cases, Excel becomes a firm’s primary operating tool,” he explained. “In contrast, Ellis seamlessly connects to the systems and documents that firms already utilize, allowing them to enhance their workflow without a complete overhaul.”

Importantly, the system maintains human engagement in the decision-making process. “Key decisions and actions still lie with human experts,” Williams emphasized. 

When asked whether he envisions a future where AI operates independently, he acknowledged, “I anticipate that the role of human involvement will decrease, but it won’t vanish entirely. Our objective is not to replace human judgment; instead, we aim to empower professionals to navigate complexities and make informed choices more efficiently.”

This article has been updated to include the addition of another investor.

Repeat founder Ryan Williams secures $10M in seed funding for an AI startup aimed at private credit managers.

A sunlit, modern startup office: Ryan Williams stands confidently at a sleek glass wall emblazoned with the “Repeat” logo, warm morning light streaming in. In front of him, a large digital display shows AI-driven credit analytics—rising bar charts titled “Private Credit Managers,” risk heatmaps, and glowing data feeds. In his hand he holds a polished golden seedling sprouting a small “$10M” tag. The scene feels optimistic and forward-looking, with natural light highlighting every detail.

Repeat founder Ryan Williams raises $10M seed for an AI startup for private credit managers - TechCrunch

- Ellis AI secures $10 million in seed funding from notable venture capital firms.  
- The company aims to streamline fragmented workflows for private credit managers using AI.  
- Founded by Ryan Williams, Ellis connects existing systems and keeps humans involved in decisions.

Ellis AI announced Thursday its emergence from stealth with $10 million in seed funding from investors including First Round Capital, 645 Ventures, Harlem Capital, Khosla Ventures, Thrive Capital, Slow Capital, Kearny Jackson, and Ariel Alternatives CEO Mellody Hobson.

Ellis uses AI agents to tackle the fragmented workflow private credit managers deal with, including managing documents, spreadsheets, and correspondence. The company was founded by Ryan Williams, best known for co-creating the real estate investment platform Cadre alongside Josh and Jared Kushner back in 2014. That company raised more than $160 million in funding and, at its peak, was valued at $800 million before being sold for an undisclosed sum to the alternative investment company Yieldstreet in 2024.

“At Cadre, I saw the next major constraint,” Williams said. “Even as the front end of private markets became more modern and accessible, the operating infrastructure underneath it remained fragmented.”

He started working on Ellis last year. The company seeks to connect and centralize all the scattered software, accounting information, and documents a private credit firm would use into one easily accessible platform. The system can flag discrepancies in the data and uses AI agents to help perform tasks like portfolio monitoring and preparing reports.

For example, Williams promises the agents can help close a fund’s books at the end of the month.

“A team may have to download files from several systems, reformat the data, compare balances, investigate discrepancies, and re-enter information by hand. In many firms, Excel becomes the operating system,” he continued. “Ellis connects to the systems and documents a firm already uses rather than forcing it to rip everything out and start over.”

It keeps a human in the loop, too, he says. “Material decisions and actions remain with the human experts,” he said.

“I expect the human loop to become narrower, but not disappear,” he continued, when asked if he sees a day when the AI works fully autonomously. “Our goal is not to replace human judgment; it’s to help people cut through the noise and make educated decisions faster.”

This piece was updated to add an investor.

Repeat founder Ryan Williams raises $10M seed for an AI startup for private credit managers

In a significant move for the tech landscape, NextSlide has officially joined forces with OpenAI, integrating its innovative team into the ChatGPT development process. Founder Ahmed Beshry shared via the NextSlide website that the company’s product enables users to transform prompts, notes, documents, or research into refined and editable presentations.

Beshry emphasized that the core mission remains focused on enhancing visual communication, striving to empower individuals to articulate their ideas with clarity. Joining OpenAI aligns with this vision, as the team will continue to create AI-driven tools that facilitate effective communication and meaningful project execution.

While the specifics of the financial arrangement have not been revealed, Beshry noted on LinkedIn that the announcement might be "a few months late," as the acquisition was finalized earlier this year. Prior to this role, Beshry co-founded Caper AI, a smart checkout solution that was acquired by Instacart in 2021.

OpenAI purchases presentation company NextSlide

A sunlit modern conference room with warm morning light streaming through tall windows; on one wall a bold NextSlide logo fills a large screen, on the opposite glass partition the words “OpenAI” appear in simple text with no logo; a sleek laptop on the table shows a slide deck morphing from NextSlide branding into OpenAI’s style, as two silhouetted executives shake hands over the deal.

OpenAI acquires presentation startup NextSlide - TechCrunch

- NextSlide has joined OpenAI, with team members now working on ChatGPT.  
- The startup aims to enhance visual communication through AI-driven presentation tools.  
- Founder Ahmed Beshry previously co-founded Caper AI, acquired by Instacart in 2021.  

In Brief

NextSlide recently announced that it’s joining OpenAI, with the presentation startup’s team members now working on ChatGPT.

The NextSlide website currently displays a note from founder Ahmed Beshry describing the startup’s product as one “that could turn prompts, notes, documents, or research into a polished, editable presentation.”

The ultimate goal, Beshry said, was “to make visual communication more accessible and help more people express their ideas clearly.” So by joining OpenAI, the team will “continue pursuing that same mission: building AI products that help people create, communicate, and turn their ideas into meaningful work.”

The financial terms of the deal were not disclosed. In a note on LinkedIn, Beshry said the announcement is coming “a few months late,” as the acquisition took place “earlier this year.”

Beshry was previously a co-founder at Caper AI, a smart cart/cashier-less checkout startup acquired by Instacart in 2021.

OpenAI acquires presentation startup NextSlide

Recent research indicates that the venture capital sector is not just a haven for questionable AI startups; it also creates an environment that can foster fraudulent activity. 

A collaborative study from the UK’s Imperial College and France’s Emlyon Business School examined 12 firms involved in 27 legal cases concerning civil or criminal allegations of securities fraud. Collectively, these cases resulted in approximately $688 million in financial losses and led to a staggering total of 73 years in prison sentences.

The researchers uncovered a consistent phenomenon referred to as “façading,” where entrepreneurs conceal the poor performance of their businesses through a range of deceptive practices. As detailed in the research, reported by TechCrunch, three distinct types of façading were identified: surface, reinforced, and deep façading. These categories reflect the degree to which there is a disparity between what audiences expect and the actual performance of the ventures.

Surface façading occurs when founders fabricate narratives of impending success—think along the lines of the AI hype that peaked in 2024, exemplified by companies like Builder.ai, which ultimately crumbled like a fragile house of cards. 

This practice intensifies into reinforced façading as entrepreneurs begin to concoct misleading documents such as bank statements and customer contracts to support their superficial claims. A notable example is iLearning Engines, a $1.5 billion AI startup that a Department of Justice investigation found was misrepresenting nearly all of its customer interactions and revenue.

The most extreme form, known as deep façading, involves extensive market manipulation, including creating fake product demonstrations, undermining internal due diligence processes, and manipulating regulations. Current leaders in the AI sector exemplify this behavior, engaging in heavy political lobbying, obstructing their internal assessments of AI safety, and often obscuring the use of human labor by dressing it up to appear as if it were autonomous AI.

The distinction between innovation in a free market and criminal activity is a fine line, especially when it involves groundbreaking technologies like AI—which currently represents a substantial $1.6 trillion gap between what investors expect and what is actually viable.

As the paper states, "By engaging in façading, entrepreneurs effectively decouple the venture’s externally projected appearance from its operational reality, making it appear to audiences as if no expectation-reality gap exists, despite the venture’s actual (subpar) performance.” It concludes that entrepreneurs cross the threshold from exaggerated storytelling to criminal deception when their narratives become increasingly detached from reality, leading to sophisticated efforts to fabricate evidence, data, and representations of their companies’ performance.

Startups Backed by VC Funding Associated with Ongoing Fraud

A moody, shadowy boardroom lit by a single dangling amber bulb. Silhouetted venture capitalists in sharp suits push heavy leather briefcases overflowing with cash across a dark wooden table. On the table lie scattered “CONFIDENTIAL” files stamped in red, legal contracts marked “FRAUD ALERT,” and a cracked smartphone screen showing a glitching startup pitch. In the background, faint outlines of tech office windows loom like silent witnesses. Dramatic low-angle shot, deep shadows and warm highlights, dark theme.

VC-Funded Startups Linked to Persistent Fraud - Futurism

- Venture capital fosters environments conducive to fraud, impacting numerous AI startups significantly.  
- Researchers identified three types of "façading" entrepreneurs use to conceal poor performance.  
- Deep façading involves extensive market manipulation and disconnects appearances from operational reality.  

Sign up to see the future, today Email address Sign Up Thank you!

It turns out the venture capital industry doesn’t just merely fund junky AI startups — it also systematically breeds the conditions for fraud.

A new joint study from researchers at the UK’s Imperial College and France’s Emlyon Business School analyzed 12 companies involved in 27 court cases related to civil or criminal cases related to securities fraud. The cases totaled around $688 million in financial losses, and resulted in 73 years of cumulative prison sentences.

After analyzing the individual cases, the researchers identified a persistent pattern of “façading,” in which startup entrepreneurs hide their flailing businesses from scrutiny through an assorted mix of fraudulent behavior.

The research, first reported by TechCrunch, identified three various types of façading: surface, reinforced, and deep façading, which are “contingent on the severity of the gap that entrepreneurs face between audiences’ performance expectations and ventures’ performance reality,” the paper explains.

The first form, surface façading, happens when startup founders create fictional stories of “imminent venture success,” like the kind of AI-washing that hit its stride in 2024, embodied by companies like Builder.ai that went on to collapse like a billion-dollar houses of cards.

This escalates into “reinforced façading” when founders begin to cook up things like bank statements or customer contracts to sell the surface story, and take pains to centralize the flow of information around the startup’s performance. One example of this might be the company iLearning Engines, a $1.5 billion AI startup which was found to be faking “virtually all its customer relationships and revenues” by a Department of Justice probe earlier this year.

Deep façading, the final and most severe form, encompasses broad forms of market manipulation, like faking product demonstrations, sabotaging internal efforts at due diligence, and manipulating regulation. This last form, one could argue, is embodied by the current front-runners of the tech industry’s AI boom, which have engaged in significant political lobbying, sabotaged their own internal research on AI safety, and routinely dress up human labor to appear like autonomous AI.

Ultimately, the line between criminality and free market innovation is thin, especially where seemingly-magical and proprietary new technologies like AI are involved (a technology which currently represents a $1.6 trillion gap between investor expectations and practical reality.)

“By engaging in façading, entrepreneurs effectively decouple the venture’s externally projected appearance from its operational reality, making it appear to audiences as if no expectation-reality gap exists, despite the venture’s actual (subpar) performance,” reads the paper. “Entrepreneurs transgress from dramatized discourse into criminal deception when their stories become unhinged from reality and involve increasingly sophisticated organizing efforts to fabricate material evidence, data, and information about their ventures’ performance.”

More on startups: We Are Mildly Horrified by This AI Startup That Coerced People Into Getting Tattoos of Its Logo in Exchange for a Job Interview

VC-Funded Startups Linked to Persistent Fraud

Jeff Dean, a pivotal figure in the development of Google's AI technologies, is embarking on a new journey by launching his own AI venture. On Wednesday, Dean, who held the title of Google's chief scientist, revealed that he, along with three colleagues, is departing the tech giant to establish Discovery Loop, a startup aimed at utilizing AI to address challenges in science and engineering.

What is Discovery Loop all about?

Discovery Loop is set up as a public benefit corporation dedicated to automating machine learning alongside scientific and engineering processes. Dean shared via X that the company's mission is to streamline the entire cycle—from the proposal of experiments to their execution and analysis—thereby accelerating advancements in various fields.

The concept of "loop" in the name refers to the company's aim to facilitate the automation of this experimental cycle. By harnessing this technology, Discovery Loop aspires to tackle some of the most complex issues in both science and engineering, potentially allowing for the parallel execution of thousands of experiments.

In its vision, Discovery Loop intends to confront 14 significant challenges identified by leading scientists and engineers to transform our way of life. These problems, once termed the "National Academy of Engineering Grand Challenges," cover topics such as "Reverse Engineer the Brain" and "Prevent Nuclear Terror." Notably, Larry Page, cofounder of Google, was part of the committee that formulated these challenges back in 2008.

Who is behind the financial backing of Discovery Loop?

In his announcement, Dean revealed that the startup has secured initial funding from esteemed venture capital firms Radical Ventures and Khosla Ventures. Participation also came from Lightspeed, Kleiner Perkins, and Doerr Capital.

Dean discussed how they created a presentation to attract a select group of venture capitalists, sharing insights from their extensive experience in the tech field. Alphabet, Google's parent company, is also a founding investor and cloud partner. Google CEO Sundar Pichai expressed gratitude for Dean's 27-year tenure at Google, acknowledging his substantial contributions to the company.

A brief overview of Jeff Dean

Jeff Dean has significantly shaped Google over the past two decades, playing a crucial role in its rise to prominence in areas like search, news, translation, and advertising. As a co-founder of the Google Brain team, he has been instrumental in transforming theoretical research in neural networks into impactful products. One of his notable contributions is the development of the Tensor Processing Unit (TPU), which has been vital in advancing AI research.

Throughout his career, Dean has collaborated with many influential figures in the AI community, including Dario Amodei from Anthropic and Ilya Sutskever from OpenAI.

Leadership team of Discovery Loop

The founding team of Discovery Loop includes Sanjay Ghemawat, Oriol Vinyals, and Quoc Le, all of whom are well-respected in their fields. Ghemawat, a longtime collaborator of Dean, shares a history that dates back before their time at Google. Together, they set a record as the first-ever level 11 engineers at Google.

Oriol Vinyals, a principal scientist at Google DeepMind, is celebrated for his work in machine learning and reinforcement learning, and he played a key role in the AlphaStar project, which bested professional players in StarCraft II.

Quoc Le, another founding member of Google Brain, has made significant strides in deep learning, including pioneering projects that utilized extensive neural networks to teach computers to recognize images.

Future implications for Google

Google has experienced significant changes, especially in the realm of AI. The acquisition of DeepMind in 2014 marked a pivotal moment for the company, as it led to innovative research, including AlphaGo, which triumphed over a seasoned board game player.

While Google laid an important foundation in AI, some former employees have propelled competitors like OpenAI to the forefront of generative AI technology, particularly with the introduction of ChatGPT in 2022. In response to evolving dynamics, Google merged its Brain and DeepMind teams and appointed Demis Hassabis as chair of Google DeepMind and chief scientist of Alphabet. Following Dean's exit, Koray Kavukcuoglu has taken over the day-to-day leadership of DeepMind, reinforcing the organization's commitment to advancing AI technologies.

5 key points to understand about Jeff Dean's new venture

A bright, sunlit modern workspace: Jeff Dean seated at a sleek wooden table strewn with notebooks, sketches and an open laptop, warm daylight pouring through floor-to-ceiling windows. Above the table hover five softly glowing icons—a rocket, a lightbulb, a neural-network node, a handshake and a shield—each representing one of his startup’s key pillars. On the wall behind him, a minimalist nameplate reads “Dean Innovations.”

5 Things to Know About Jeff Dean's New Startup - Business Insider

- Jeff Dean, Google's chief scientist, co-founded AI startup Discovery Loop focused on automation.  
- Discovery Loop aims to address 14 significant challenges in science and engineering through AI.  
- Initial funding comes from notable investors including Radical Ventures and Alphabet as a partner.

Legendary Googler Jeff Dean is starting his own AI company.

On Wednesday, Dean, Google's chief scientist, announced that he and three other colleagues were leaving to start Discovery Loop, an AI startup focused on AI applications in science and engineering.

Here's everything we know about the new venture

What is Discovery Loop?

Discovery Loop is a public benefit corporation "whose mission is to automate machine learning, science, and engineering to accelerate discoveries and progress," Dean wrote on X.

The loop part of the name comes from the company's goal of automating the process by which scientists and engineers propose an experiment, implement and run it, and evaluate the results.

Related video LinkedIn marketing leader Jessica Jensen said flexibility will help companies attract and retain top talent

The company hopes this automation can unlock some of the trickiest problems in science and engineering by enabling "parallel execution of thousands of experiments."

Introducing Discovery Loop ♾. We’re building AI to run experiments at unprecedented scale and to solve the biggest bottlenecks in science and engineering.



Our founders @JeffDean, @Sanjay_Ghemawat, @quocleix, @OriolVinyalsML helped build the AI infrastructure and models that… pic.twitter.com/Y3FUlMLTbf — Discovery Loop (@DiscoLoopAI) August 5, 2026

Specifically, Discovery Loop wants to build systems that can tackle parts of the 14 challenges laid out by some of the world's most accomplished scientists and engineers to improve life as we know it. Formerly known as the "National Academy of Engineering Grand Challenges," the 14 topics range from "Reverse Engineer the Brain" to "Prevent Nuclear Terror."

Interestingly, the 2008 committee that came up with the grand challenges included Larry Page, the cofounder of Google.

Who is funding Discovery Loop?

According to Dean, Radical Ventures and Khosla Ventures, two Silicon Valley mainstays, led the initial round. Lightspeed, Kleiner Perkins, and Doerr Capital also participated.

We created a pitch deck to tell a handful of VC firms about us and what we were up to (a fun experience!). Here’s a few slides about our background and some of the things we’ve worked on from the pitch deck (it was fun putting together the list of people in our teams who have… pic.twitter.com/7glcm9T9XG — Jeff Dean (@JeffDean) August 5, 2026

In a sign of how he left things at Google, Alphabet is a founding investor and cloud partner, Google CEO Sundar Pichai said.

"I also want to give a huge thanks to the incomparable @JeffDean after an incredible 27-year run at Google," Pichai wrote on X.

Who is Jeff Dean?

Jeff Dean Bloomberg/Getty Images

If you've used Google in the last two decades, you have Jeff Dean to thank. As Google's 30th employee, his tenure spans from establishing Google's dominance in search to the creation of generative AI models.

Dean helped advance Google Search, Google News, Google Translate, and Google Ads.

If that wasn't enough, Dean co-founded the Google Brain team that, as the "Godfather of AI" and Nobel laureate Geoffrey Hinton put it, turned "basic research in neural nets into game-changing products." One of the biggest breakthroughs Dean helped develop was the Tensor Processing Unit (TPU), which is credited with accelerating AI research.

As Google's AI head, Dean oversaw AI research and later co-led the development of Gemini, Google's family of models.

Many of the biggest names in AI worked alongside Dean, including Anthropic CEO Dario Amodei, OpenAI cofounder Ilya Sutskever, and more.

Who else is leading Discovery Loop?

Announcing Discovery Loop!



I am very excited to announce that, along with my longtime friends and collaborators @Sanjay_Ghemawat, @OriolVinyalsML and @quocleix, we are founding Discovery Loop (@DiscoLoopAI), a Public Benefit Corporation whose mission is to automate machine… pic.twitter.com/ancoplyNvN — Jeff Dean (@JeffDean) August 5, 2026

Sanjay Ghemawat: Ghemawat is one of Dean's oldest collaborators, going back to before the pair joined Google when they worked at Digital Equipment Corporation.

At Google, the pair became so inseparable that Dean's kids called Ghemawat "Uncle Sanjay," according to a 2018 profile in The New Yorker. Ghemawat and Dean also became the first-ever level 11 Google engineers, carrying the title of Senior Fellows.

According to Dean's slide deck, he is the most-cited distributed systems researcher; Ghemawat is the fourth most cited.

Oriol Vinyals: Vinyals, a principal scientist at Google DeepMind, is also joining Discovery Loop.

According to his Google biography, Vinyals focused on machine learning and reinforcement learning. Before joining the team, he worked on projects that fuel tools like Google Translate.

One of his research projects, called AlphaStar, was an agent that bested a top-tier professional StarCraft II player.

Among the most-cited AI researchers, per Dean's slide deck, Vinyals is sixth.

Quoc Le: Le was a founding member of the Google Brain team. At Stanford, his advisor was Andrew Ng, who helped establish the Brain team.

Le worked on one of Google Brain's first big projects: developing a neural network of 1,000 machines containing 16,000 processors that demonstrated machine learning by teaching itself how to recognize items after being trained on 10 million images randomly pulled from YouTube videos, as The New York Times reported in 2012.

This being the internet, the system taught itself to recognize cats.

What's next for Google?

Demis Hassabis, CEO of DeepMind Shane Anthony Sinclair/Getty Images for Cannes Lions

Google is no stranger to big shake-ups, including in AI.

In 2014, the search giant acquired DeepMind, then a startup. Under Google's banner, DeepMind conducted influential research, including AlphaGo, a program capable of beating a master at the challenging board game Go.

Despite laying the foundation for several AI breakthroughs, Google saw some of its former employees, including Sutskever, push OpenAI to the front of the generative AI race with the release of ChatGPT in 2022.

In 2023, Google merged the Brain and DeepMind teams into one group led by Demis Hassabis.

In addition to Dean's departure, Google announced that Hassabis would become chair of Google DeepMind and chief scientist of Alphabet.

Koray Kavukcuoglu, chief technology officer of Google DeepMind and Google's chief AI architect, assumed the day-to-day operations of DeepMind and the title of senior vice president.

5 things to know about Jeff Dean's new startup

A Seattle-based educational technology startup, Wild Zebra, has successfully secured $6 million in funding to enhance its AI-driven platform designed for teaching math and reading. This funding follows promising initial results indicating that Wild Zebra’s innovative approach can compete effectively with the free educational resources recently launched by major players such as OpenAI, Google, and Anthropic.

What sets Wild Zebra apart is its Socratic method of instruction. Instead of providing direct answers, the AI engages students by guiding them toward the solution through a series of thought-provoking questions, thereby encouraging critical thinking and self-discovery.

Additionally, the lessons on Wild Zebra are tailored to align with students' personal interests, whether that be sports, cooking, music, or other hobbies. This personalization aims to make the learning process more relatable and engaging.

The platform also monitors interactions to pinpoint not just disengagement or distractions but also to celebrate moments of persistence and curiosity, providing valuable insights for teachers and parents through dashboards. “We want to recognize when kids are exhibiting positive behaviors as well,” said Edan Shahar, co-founder and CEO of Wild Zebra.

The recent funding, raised through a notably oversubscribed seed round, was spearheaded by Trilogy Equity Partners with support from Tetherpoint Capital and notable angel investors, including Shrikesh Majithia. This latest investment increases the company's total funding to $8 million.

Co-founded in 2024 by Shahar and AI expert Erik Selberg, Wild Zebra caters to students from grades 2 to 9. With a current team of ten, the startup plans to utilize the new funds primarily for hiring engineers and bolstering sales and marketing strategies.

Amy McCullough, Managing Director at Trilogy and a new board member at Wild Zebra, emphasized the potential for the platform to provide parents and educators with a real-time understanding of each student’s progress and a clear roadmap for their learning journey. McCullough, who initially came to know Shahar through his successful Test Innovators company, was drawn to Wild Zebra’s potential for tracking students’ learning over time.

When Wild Zebra was first highlighted by GeekWire last year, competitors like OpenAI and Google had just unveiled their own educational tools. Recently, Anthropic released a free version of its AI, Claude, tailored for K-12 teachers, while Khan Academy’s own AI, Khanmigo, has been in use for several years.

Shahar commented on the limitations of general-purpose chatbots, which can effectively answer individual questions but lack the ability to track a student's learning progression. In contrast, Wild Zebra employs a unique “learning tree” approach that monitors what a student has achieved, identifies gaps, and determines future learning objectives.

As for the competitive landscape, Shahar views the prominent AI labs more as collaborators than adversaries, leveraging their models within Wild Zebra’s framework.

Currently, the startup has expanded its reach to tens of thousands of students, up significantly from 6,000 just a year ago, and has facilitated hundreds of thousands of tutoring sessions. Families can subscribe to the platform for $48 per month per child, with initial users primarily being parents from pilot schools eager to enroll their other children.

On the institutional side, Wild Zebra collaborates with E3n, an organization formed from the merger of the Educational Records Bureau and the Enrollment Management Association, which serves selective private schools. The ERB invested in Wild Zebra in 2024, utilizing its exam data to determine each student’s starting level, although not all partnering schools do so.

Preliminary evaluations conducted by E3n with a sample of 722 students have shown notable advancements in math and reading skills, particularly among fifth and sixth graders, boasting gains of four to eight percentage points compared to ERB benchmarks.

While the platform initially focused on private schools, Wild Zebra is now exploring partnerships with charter and public institutions. Shahar noted that schools and students skeptical about incorporating AI tend to shy away from their services, while those who opt in usually share a positive outlook, albeit with an awareness of potential risks.

Among the most common concerns voiced by students is the environmental impact of AI technologies. Shahar admitted that while he acknowledges the risks associated with AI, he remains concentrated on exploring its positive potential. “I am not overly optimistic; I recognize the challenges of AI, but my work is focused on maximizing opportunities,” he said.

This startup has secured $6M for an AI tutor designed to help children learn independently.

A bright, sunlit study nook by a large window: on a wooden desk sit open textbooks, a steaming mug, a small potted plant, and a laptop screen showing a friendly AI tutor interface of speech bubbles and lightbulb icons. Beside it, a neatly wrapped stack of cash marked “$6M” hints at fresh funding. Warm natural light floods the scene.

This startup just raised $6M for an AI tutor that helps kids figure it out themselves - GeekWire

- Wild Zebra, an edtech startup, raised $6 million to enhance its AI learning platform.  
- The platform employs a Socratic method, guiding students through questions rather than direct answers.  
- Customized lessons engage students by incorporating their interests, promoting engagement and persistence.

This startup just raised $6M for an AI tutor that helps kids figure it out themselves

A Wild Zebra math problem framed around baking, one of the interests the student selected. After the student works out that each section is 1/12 of the cookie, the AI confirms the step and asks the next question rather than finishing the problem. (Wild Zebra images, click to enlarge)

Seattle-based edtech startup Wild Zebra has raised $6 million to expand its AI learning platform for math and reading, citing early signs that its specialized product can hold its own against the free study tools released over the past year by OpenAI, Google and Anthropic.

One of Wild Zebra’s secrets is its Socratic approach: rather than handing over the answer, the platform works students toward it with questions, drawing out the reasoning step by step.

Another key difference: the lessons are customized to be grounded in whatever topics or hobbies the student is already interested in, such as sports, cooking, music, or anything else.

The platform also analyzes the conversations themselves, looking not just for students who are distracted or getting answers elsewhere, but for those showing persistence or curiosity — flagging those positive moments in dashboards for teachers and parents.

The idea is “to catch kids being good, too,” said edtech veteran Edan Shahar, the company’s CEO and co-founder.

Latest funding: The oversubscribed seed round was led by Bellevue, Wash.-based Trilogy Equity Partners, with participation from Tetherpoint Capital and angel investors including Shrikesh Majithia. It brings the company’s total funding to $8 million.

Wild Zebra co-founders: CEO Edan Shahar, left, and CTO Erik Selberg. (Wild Zebra Photos)

Wild Zebra, co-founded in 2024 by Shahar and longtime AI technologist Erik Selberg, serves students in grades 2 through 9. The company, with a team of 10, plans to use the new funding for hiring, primarily engineers, along with sales and marketing initiatives.

Trilogy Managing Director Amy McCullough, who is joining the Wild Zebra board, said the VC firm backed the company for its potential to give parents and teachers a real-time view of what a student has actually learned, and a path to mastering the material from there.

She said she came to the deal as a customer first: her family had used Test Innovators, the private-school admissions test-prep company Shahar ran for nearly a decade, and she was struck by how well it delivered for parents and students at a high-stakes moment.

“When we were introduced to Wild Zebra, we knew very quickly that this was the team who could build the student-centric AI partner for the market,” McCullough said.

Competitive landscape: When GeekWire first covered Wild Zebra last year, OpenAI and Google had just launched study modes of their own. Anthropic launched a free version of Claude for K-12 teachers last month, and Khan Academy’s Khanmigo has been in the market for years.

Shahar said the general-purpose chatbots are good at answering a single question or explaining a concept, but don’t keep track of what a student knows over time.

Wild Zebra builds what the company calls a “learning tree” for each student — what they’ve mastered, where the gaps are, what comes next — and uses it to decide what they work on next, rather than just responding to whatever they happen to ask.

Shahar sees the big AI labs as suppliers more than rivals. Wild Zebra runs on their models, along with others.

Traction so far: Wild Zebra says it now reaches tens of thousands of students, up from about 6,000 a year ago, and has logged hundreds of thousands of tutoring conversations.

It opened the platform to families this year at $48 per month per child. The first consumer customers, Shahar said, were parents at pilot schools who wanted it for their other kids.

On the school side, Wild Zebra works through E3n, formed this year by the merger of the Educational Records Bureau and the Enrollment Management Association, which serve selective private schools. ERB invested in the company in 2024.

Many member schools test students using E3n’s exams, and Wild Zebra uses those results to set each student’s starting point, though some schools using the platform don’t.

Early results: E3n also conducted a 722-student pilot evaluation of math and reading, measuring results against ERB norm groups collectively built from hundreds of thousands of students. It found gains of four to eight percentage points among fifth and sixth graders.

Having started in private schools, Wild Zebra is now talking with charter and public schools.

Shahar said schools and students skeptical of AI mostly don’t become customers, and the ones who do tend to share his optimism while still taking the risks seriously.

The concern he hears most often from students is AI’s environmental impact — which, he acknowledged, is not something Wild Zebra is working to address.

“I’m not a Pollyanna. I think there are certainly potential downsides to AI in general,” he said. “There are people working on the downside mitigation. I’m working on upside opportunity.”

This startup just raised $6M for an AI tutor that helps kids figure it out themselves

Google announced a significant shake-up in its AI leadership on Wednesday, as the company’s chief scientist, Jeff Dean, departs after a remarkable 27 years with the tech giant. This exit is part of a broader trend of prominent figures leaving Google as the company strives to advance its position in the competitive realm of artificial intelligence.

Demis Hassabis, CEO of Google DeepMind, is stepping into the role of chairman and chief scientist. Koray Kavukcuoglu, the technology head at Google DeepMind, will take charge of developing the new Gemini 4 model, a transition outlined by CEO Sundar Pichai in a recent post on X.

According to a blog update, Dean is exiting to embark on a new venture, Discovery Loop, an AI startup he is co-founding with three other seasoned colleagues from Google. Pichai commended Dean for his exceptional contributions over nearly three decades, affirming that he leaves on amicable terms. Furthermore, Alphabet, Google’s parent company, plans to invest in Discovery Loop and will offer cloud services and computational resources to support the startup’s endeavors.

This leadership change comes amid ongoing challenges for Google as it seeks to reassure investors regarding its substantial investments in AI technology, particularly as competitors like OpenAI with ChatGPT and Anthropic with Claude remain dominant in the chatbot sector.

Notably, several high-profile engineers have departed from Google recently. Noam Shazeer, a distinguished computer scientist, has joined OpenAI, while John Jumper, awarded a Nobel Prize for his development of the AlphaFold AI model, has moved to Anthropic.

In the most recent financial quarter, Google revealed it experienced a negative cash flow for the first time, forecasting total capital expenditures could reach as high as $205 billion this year, despite not providing updates on the release of the new Gemini 3.5 Pro model. Consequently, Alphabet’s shares dipped by 4.2% on Wednesday, prompting concerns among investors about the company's pace in the AI sector.

Pichai emphasized the company’s commitment to leading the field, stating in his blog post, “We are dedicated to pushing the boundaries and intensely focused on areas needing improvement. I’m genuinely thrilled about our forthcoming model releases and the advancements we’re witnessing. We need to maintain our momentum and clarity of purpose moving forward.”

In addition to his new role, Hassabis will continue overseeing Isomorphic Labs, a London-based AI drug discovery firm that originated from Google. As the co-founder of DeepMind—acquired by Google in 2014—Hassabis has been pivotal in shaping Alphabet’s AI initiatives.

Dean's new enterprise, Discovery Loop, intends to streamline the scientific discovery process, which he contends is currently inefficient and labor-intensive. His vision includes harnessing AI for faster and more effective discoveries, particularly in drug development and tackling global issues like sustainable energy.

The founding team for Discovery Loop also comprises Oriol Vinyals, Quoc Le, and Sanjay Ghemawat, some of the foremost researchers in the scientific community. Alongside Alphabet's investment, Radical Ventures and Khosla Ventures are collaboratively spearheading the startup's seed funding round.

Additionally, Alphabet maintains a multi-year partnership with News Corp, the parent company of The Post, aimed at advancing AI-focused content and products.

Google's chief scientist Jeff Dean, who has served for 27 years, is departing amid a restructuring of AI leadership to join a startup.

A soft golden dawn bathes the entrance of Google’s sprawling campus, a large, vibrant Google logo mounted on the gate. A lone figure evoking Jeff Dean, briefcase in hand, strides away from a small plaque reading “Chief Scientist – 27 Years” toward a sleek, glass-fronted building in the distance labeled “New Venture.” Natural morning light, cinematic composition.

Google's Jeff Dean -- chief scientist for 27 years -- leaving during AI leadership overhaul for startup - New York Post

- Google announced significant AI leadership changes as chief scientist Jeff Dean departs.  
- Demis Hassabis becomes chairman and chief scientist, focusing on Google’s AI developments.  
- Alphabet will invest in Dean's new startup, Discovery Loop, aimed at automating discovery.  

See more of our coverage in your search results.

Google on Wednesday revealed an overhaul of its AI leadership, as its chief scientist for nearly three decades exits – the latest in a string of high-profile departures to plague the tech giant in its struggle to pull ahead in the artificial intelligence race.

Demis Hassabis, chief executive of Google DeepMind, will become the company’s chairman and chief scientist, while Google DeepMind’s technology chief, Koray Kavukcuoglu, will lead work on Google’s new Gemini 4 model, Google CEO Sundar Pichai announced in a post on X.

Jeff Dean – Google’s current chief scientist and one of the company’s first employees, working there for 27 years – is leaving to start his own AI startup, Discovery Loop, with three other company veterans, according to a company blog post.

Advertisement

3 Demis Hassabis, chief executive of Google DeepMind, will become the company’s chairman and chief scientist. via REUTERS

Pichai said Dean is leaving the company on good terms, congratulating him for “an incredible 27-year run.” Google’s owner, Alphabet, will invest in Discovery Loop and provide the startup with cloud and computing capacity, Pichai added.

It’s just the latest exit from Google’s tech leadership as the Silicon Valley giant attempts to convince investors that its massive investments in AI will pay off – especially as OpenAI’s ChatGPT and Anthropic’s Claude models remain firmly in the lead for the top chatbots.

Advertisement

Noam Shazeer, a longtime Google computer scientist and leading AI researcher, recently left the company for rival OpenAI, while John Jumper, who won a Nobel Prize for his work on an AlphaFold AI model, jumped ship to Anthropic.

Meanwhile, Google reported that it turned cash flow negative for the first time in the most recent quarter while forecasting full-year capex of as much as $205 billion – even as it failed to share an update on the release of its newest Gemini 3.5 Pro model.

Shares of Alphabet fell 4.2% Wednesday as investors fear the legacy tech company is losing steam in the AI race.

“We are committed to being at the frontier, and are super focused on the areas where we need to improve,” Pichai said in Wednesday’s blog post. “I’m really excited for our upcoming model releases and the progress we’re seeing. We have to continue to move fast and with clear purpose here.”

Advertisement

Along with moving into the chairman role, Hassabis will continue running Isomorphic Labs, a London-based AI drug discovery startup that was spun off from Google.

3 Jeff Dean, Google’s current chief scientist, is leaving to start his own AI startup. AFP via Getty Images

Hassabis is the co-founder of DeepMind, the AI research lab that Google purchased in 2014. He has been leading Alphabet’s AI divisions over the past few years.

Dean’s new startup, Discovery Loop, aims to automate scientific discovery, arguing the current manual process is slow and labor-intensive and can be made more efficient with the use of AI.

Advertisement

Longtime Google workers Oriol Vinyals, Quoc Le and Sanjay Ghemawat – some of the world’s top scientific researchers – are joining Dean to run the new venture.

Its goal is to scale this discovery automation to help advance efforts in drug discovery and world challenges like clean energy.

3 Alphabet will be investing in Discovery Loop and providing the startup with cloud and computing capacity. REUTERS

Along with an investment from Google, Radical Ventures and Khosla Ventures are co-leading a seed round for the new company.

Alphabet has a multi-year partnership with News Corp, which owns The Post, to develop AI-related content and products.

Google’s Jeff Dean — chief scientist for 27 years — leaving during AI leadership overhaul for startup

Jeff Dean, a prominent figure at Google with a long tenure, is departing from the company to embark on an entrepreneurial journey focused on artificial intelligence. He is joined by a team of distinguished researchers, including Sanjay Ghemawat, a leading engineer and senior fellow at Google; Quoc Le, a pivotal AI researcher and co-founder of Google Brain; and Oriol Vinyals, a senior research scientist from Google DeepMind. Dean aims to take on the role of CEO for their new venture.

The new endeavor is called Discovery Loop, envisioned as a public benefit corporation dedicated to harnessing AI to enhance scientific research. The organization intends to deploy sophisticated algorithms to run thousands of experiments concurrently, with the ambition of partially automating research processes and significantly increasing the scale of experimentation.

One of the innovative concepts Discovery Loop plans to explore is using AI to aid in the development of more advanced AI systems, a practice termed recursive self-improvement, which would ultimately remove the need for human input in the iterative process.

According to the company's press release, “Although science and engineering have advanced society tremendously over the years, innovation has historically been hindered by slow, sequential human iterations that create a notable bottleneck.” Discovery Loop is striving to create AI systems capable of leveraging vast computational power to drastically enhance the speed and efficiency of innovation through automation of entire experimental processes.

The integration of AI in speeding up scientific discoveries has intrigued both scientific and technological spheres for years, but only recently has it shifted from being predominantly experimental to having viable commercial opportunities.

Discovery Loop has secured funding from various sources, including Alphabet, Google's parent company. The startup's initial funding round is co-led by Radical Ventures and Khosla Ventures, with contributions from Kleiner Perkins, Lightspeed, and Doerr Capital.

The founding team stated, “The next significant frontier for AI is to transition from merely answering questions to actively making discoveries. By fundamentally speeding up how engineering and scientific exploration occurs, we can introduce the benefits of groundbreaking technologies to the world much more rapidly.”

Having joined Google in 1999 as the company's 30th employee, Dean has been instrumental in developing critical components of Google’s search infrastructure, including its systems for crawling, indexing, and query processing. He has also played a vital role in the advancement of Google Gemini's multimodal models and was central to the company's early efforts in AI research.

Dean expressed his vision in an interview with the New York Times, stating, “We see an opportunity for AI to fully automate what has traditionally been a very human-centered experimental loop. This approach will not only yield a greater quantity of experiments but also improve their quality, ultimately leading to significant scientific breakthroughs and advancements.”

Jeff Dean and several leading AI researchers are departing Google to create their own startup.

A sunrise-lit, ultra-modern glass building with a bold Google logo on its façade. In the foreground, a small group of professional researchers in business casual stride confidently away from the campus toward a warm, glowing horizon. One carries a sleek tablet showing bright AI neural-network schematics—symbolizing their new startup. Soft natural light, hopeful and forward-looking mood.

Jeff Dean and other top AI researchers are leaving Google to launch their own startup - TechCrunch

- Jeff Dean is leaving Google to co-found an AI startup called Discovery Loop.  
- Discovery Loop aims to automate scientific research using advanced AI algorithms and technology.  
- The startup has received backing from Alphabet and notable venture capital firms for funding.  

Jeff Dean, one of the longest-serving and most influential executives at Google, is stepping down from the search giant to launch his own AI startup.

Coming with him as co-founders are several other top researchers at the company, including Sanjay Ghemawat, a top engineer and senior fellow at Google; Quoc Le, a key AI researcher and founding member of Google Brain; and Oriol Vinyals, a senior research scientist at Google DeepMind. Dean reportedly plans to serve as CEO.

Together, the group is starting Discovery Loop, a public benefit corporation that seeks to use AI to turbo-charge scientific research.

Discovery Loop says it plans to use high-octane algorithms to initiate and iterate thousands of experiments simultaneously, with the goal of partially automating the research process and expanding the scale at which experimentation can be conducted.

The startup is also interested in using AI to help create more powerful AI (a process known as recursive self-improvement), which would cut human iteration out of the loop entirely.

“While science and engineering have tremendously advanced society over past centuries, progress has traditionally relied on slow, sequential human iterations, creating a significant bottleneck,” the company said in a press release. “Discovery Loop is developing advanced AI systems that leverage massive computational scale to fundamentally transform the speed and efficiency of innovation by automating complete experimental loops.”

Using AI to accelerate scientific discovery has been a key interest of the science and tech communities for many years, but until recently, it remained a largely experimental field with limited commercial application.

The company has received financial support from a number of different sources, including Google’s parent company Alphabet. The initial funding round is being co-led by Radical Ventures and Khosla Ventures, the company announced Wednesday. Kleiner Perkins, Lightspeed, and Doerr Capital also participated.

“The next great frontier for AI is to go beyond answering questions and to begin making discoveries,” the founding team said in a joint statement. “By fundamentally accelerating how engineering and scientific discovery are conducted, we can deliver the benefits of transformative technologies to the world far sooner.”

Dean has worked at Google since 1999 and was Google’s 30th employee. Over the years, he has contributed to Google search’s core infrastructure, including its crawling and indexing system and its query-serving system. He has also had a significant impact on Google Gemini’s multimodal models and played a leading role in the company’s early AI research.

“We think there is opportunity for AI to more fully automate what has traditionally been a very human-intensive experimental loop,” Dean told the New York Times. “You will get both a higher quantity and a higher quality of experiments, and that will lead to scientific breakthroughs and advances.”

Jeff Dean and other top AI researchers are leaving Google to launch their own startup

On July 25, renowned engineer Jeff Dean addressed an audience of 6,000 aspiring entrepreneurs at Y Combinator's Startup School, held at the Chase Center in San Francisco. Dean, who has made significant contributions to Google as a cofounder of Google Brain and the chief scientist at Google DeepMind and Google Research, is also the technical colead of the company’s premier AI initiative, Gemini. He has collaborated extensively with Sanjay Ghemawat, developing critical components of Google’s computing and search systems. A humorous anecdote circulated in tech circles suggests that his resume lists only what he hasn’t accomplished, as that would be considerably shorter. During his talk, Dean was asked about essential areas for startup founders to consider.

Dean emphasized the necessity of selecting a project that one is passionate about and that holds real-world relevance. He shared his personal interest in creating an automated version of the scientific method. He described the approach: “You propose an experiment, set up the necessary elements to conduct it, assess its outcomes, and then analyze the results.” He noted that executing thousands of such automated experiments could lead to revolutionary advancements in fields ranging from science and biology to chip design and the enhancement of AI models.

What the audience didn't realize was that Dean was unveiling his own covert startup, which he had recently launched. Today, this venture is official: after nearly 27 years with Google, Dean, along with Ghemawat and two other elite AI researchers, is embarking on a new journey with a company called Discovery Loop.

While Google will maintain a stake in this new endeavor, the departure is a significant setback for the tech giant, which is striving to remain competitive in the rapidly evolving AI landscape. The exit of Dean and Ghemawat, who were among Google’s inaugural hires, is akin to legendary musicians like Mick Jagger and Keith Richards leaving the Rolling Stones to venture out on their own. To compound matters, the remaining cofounders bring their own astounding credentials: Oriol Vinyals, a VP of research at DeepMind and a technical leader for Gemini, and Quoc Le, a cofounder of Google Brain noted for his work on Google’s AutoML-Zero project, which employs machine learning algorithms to autonomously develop AI tools.

In an interview I conducted with the cofounders just ahead of their launch, Dean revealed that the concept for Discovery Loop emerged only a few weeks prior. He was quickly joined by his three collaborators, who not only share a professional history but also friendship, often vacationing together. “Though we were each working on different projects, we began to recognize how AI could potentially automate scientific and engineering processes,” he recalled.

The team's initial plan outlines that Discovery Loop will serve as its own first client, with the autonomous experimentation loops being used to refine the company’s advanced machine learning algorithms. This enhanced software will lay the groundwork for an AI system capable of addressing challenges across various domains. Quoc Le expressed his enthusiasm: “I’m very excited about automating machine learning. It might even lead us to discover a novel transformer architecture.”

Four Leading AI Experts from Google Are Departing to Start Their Own AI Company

A dark city street at dusk outside Google’s towering glass HQ, the giant, illuminated Google logo dominating the facade. Four professional silhouettes stride out of the revolving doors, each cradling a pulsing, neuron-web sphere of light. Across the street stands a smaller, sleek startup loft, its windows aglow with warm golden light hinting at fresh innovation. Deep shadows stretch behind the figures, amber streetlamps casting a moody glow as they depart to launch their own AI venture.

4 of Google’s Top AI Brains Are Leaving—and Launching Their Own AI Startup - WIRED

- Jeff Dean announced his departure from Google to cofound Discovery Loop startup.  
- Discovery Loop aims to automate the scientific method for advancements in AI and engineering.  
- Google will retain a stake in Discovery Loop, posing challenges for the search giant.  

On July 25, Jeff Dean spoke to 6,000 would-be founders at Y Combinator’s Startup School at San Francisco’s Chase Center. Dean is a legendary engineer at Google. He is cofounder of Google Brain, chief scientist of Google DeepMind and Google Research, and technical colead on the company’s flagship AI model Gemini. With his frequent collaborator Sanjay Ghemawat, he’s written significant parts of the company’s computing and search infrastructure. A running joke in geekdom is a list of surreal “facts” about his technical prowess: Jeff Dean’s resume lists the things he hasn’t done; it’s shorter that way. At Chase Center, the interviewer asked him what areas startup founders should think about tackling.

Dean said that the most important thing would be to pick something you’re excited about and would be useful in the world. He then shared his own interest: an automated version of the scientific method. “You propose an experiment, you implement what you need to run the experiment, you evaluate the experiment, and then you get results from that,” he said. Dean said that running thousands of those automated loops could unlock groundbreaking advances in multiple domains like science, biology, chip design, and even making improved AI models.

Unbeknownst to the crowd, Dean was talking about his own secret startup, which he had just organized. Today it’s official: After almost 27 years, Dean is leaving Google, along with Ghemawat and two other top-tier AI scientists, to found a company called Discovery Loop.

Even though Google will take a stake in the new company, it’s a devastating blow to the search giant as it attempts to keep pace in the frantic AI model competition. It’s bad enough that Dean and Ghemawat, who were among the company’s first hires, are leaving—that’s like Mick Jagger and Keith Richards ditching the Rolling Stones to start a new band. Worse, the other cofounders are also supernovas in the AI firmament: Oriol Vinyals, VP of research at DeepMind and a technical lead for Gemini, and Quoc Le, a cofounder of Google Brain and the key scientist behind Google’s AutoML-Zero, a project that uses machine learning algorithms to autonomously build AI products.

In an interview I did with the cofounders just before launch, Dean tells me the idea came up only a few weeks ago. He was quickly joined by the other three, who are not only long-time colleagues but friends who vacation together. “We were working on different things, but we were all starting to see the possibility of AI being able to automate scientific and engineering loops,” he says.

The quartet worked up a road map wherein Discovery Loop would be its own first customer. The autonomous experiment loops they create initially will be directed towards improving the company’s advanced machine learning algorithms. That improved software will then become the core of an AI system that can tackle problems in subsequent domains. “I’m very excited about automating machine learning,” says Le. “It might be that we will discover a different transformer architecture.”

4 of Google’s Top AI Brains Are Leaving—and Launching Their Own AI Startup

Google wants its AI to get better at coding. It might have found a shortcut.

The tech giant has been in discussions with San Francisco startup Mechanize in recent weeks for a potential deal that would involve Google hiring some of Mechanize's talent, four people familiar with the conversations said.

The deal, which some of the people said is worth over $1.5 billion, is in progress, and details could change. Google is discussing a non-exclusive licensing agreement for Mechanize's technology as part of the deal, one person familiar with the matter said. The talent Google could acquire from Mechanize would work on model evaluation and development, the person added.

The talks highlight two realities of the AI boom. Coding has become one of the most important — and lucrative — applications of AI, and Big Tech companies are getting creative about how they acquire the talent and technology they need to stay competitive.

It wouldn't be the first time Google has done a workaround deal for talent and technology. In the past couple of years, Google has structured acquisitions as hybrid transactions that include bringing in talent via acquihires and gathering technology through licensing and other methods. Companies sometimes take this approach to avoid the antitrust scrutiny of full acquisitions.

Related video Microsoft is betting big on AI. Company insiders have serious doubts.

Last year, Google swept in to acquire Windsurf's talent and licensed its tech after OpenAI tried to buy it. Windsurf's CEO, Varun Mohan, now leads Google's Antigravity, an agentic coding platform. In 2024, the search giant rehired Character AI cofounder Noam Shazeer and paid for non-exclusive rights to use the startup's AI technology (Shazeer recently left the company to join OpenAI).

Google declined to comment over email. Mechanize declined to comment.

Mechanize launched last year with a mission of automating every job and a star-studded group of investors, including former GitHub CEO Nat Friedman, Stripe CEO Patrick Collison, and podcaster Dwarkesh Patel.

The startup said earlier this year that it raised $9.1 million in a funding round at a $500 million valuation. Mechanize's CEO, Tamay Besiroglu, previously cofounded Epoch AI, which also focused on testing AI models.

Mechanize's tech can help tech companies improve the performance of their AI models at coding — something Google has struggled with, while OpenAI and Anthropic have scooped up developer customers with Codex and Claude Code.

Mechanize has wider ambitions than coding. "Our current focus is software engineering, but our long-term goal is the full automation of valuable work across the economy," its website reads.

Google is in talks for a $1.5 billion-plus deal with AI coding agent startup Mechanize

Generate a cinematic news thumbnail for title:Google is in talks for a $1.5 billion-plus deal with AI coding agent startup Mechanize. No text in image.

Google is in talks for a $1.5 billion-plus deal with AI coding agent startup Mechanize - Business Insider

When sales representatives from major data-labeling firms in Silicon Valley attended the International Conference on Machine Learning in Seoul this year, they were geared up to attract significant clients within the AI sector. These companies, collectively valued at tens of billions and generating substantial profits by supplying training data to organizations such as OpenAI and Anthropic, are accustomed to navigating the demands of the often unpredictable AI labs.

However, they discovered another potential market thriving on their doorstep: China’s rapidly advancing AI sector.

Several Chinese firms have laid out their acquisition needs. Tencent, a corporation identified by the U.S. as having ties to the military—an assertion they contest—distributed a comprehensive request for training data that includes areas such as finance and cybersecurity, as well as ambitious objectives like developing AI systems that can enhance their own learning processes.

While the U.S. has long restricted China's access to the advanced chips that drive premier AI models, it has not imposed similar limitations on the data necessary for training these systems. This expertly curated human knowledge, developed through specialized networks and shaped by detailed task specifications, serves as the backbone of the data infrastructure that trains models to handle sophisticated tasks, such as financial analysis and software development. This lucrative industry represents hundreds of millions of dollars each year and is contributing to the relentless pursuit of technological parity between Chinese and American AI models.

Nathan Lambert, previously a research scientist at the Allen Institute for AI, stated, “After the compute needed to train the models, quality data is paramount. As we aim for models to tackle new and more complex areas, securing high-quality data becomes the most impactful factor in making a model effective.”

Forbes has reviewed communications indicating a burgeoning trade worth several hundred million dollars in training datasets between American data firms and China's foremost AI laboratories. The same startups servicing OpenAI and Anthropic are also supporting top Chinese labs, effectively putting Silicon Valley in a position where it is fueling both sides of the evolving AI rivalry.

In the quest to match American AI capabilities, Chinese labs can pursue three main strategies: recruiting top talent from the U.S., distilling insights from outputs generated by systems like ChatGPT or Claude (which technically contradicts their terms of service), and more straightforwardly, acquiring the same training data from the same suppliers as American labs.

Training an AI to master intricate tasks, such as complex accounting or experimental protocols, goes beyond mere data collection. The real advantage lies in the design of the data itself, including which data types to emphasize and the sophisticated criteria crafted by experts to guide the model’s learning process. By purchasing these proprietary datasets, Chinese labs efficiently acquire the extensive expertise that fuels Silicon Valley’s advanced models.

Sean Cai, an AI data consultant and blogger, summed up the situation: “Analyzing the data landscape in China and the overall supply chain reveals that a considerable amount of U.S. data facilitating advancements in the U.S. is also being sold to Chinese labs.”

The shadowy data supply chain in China

This trade network channels directly from American data-labelling companies to leading Chinese technology giants like Tencent. In one correspondence reviewed by Forbes, a Tencent data procurement officer mentioned that they partner with Surge AI, which has provided data to the likes of the U.S. Army and Air Force. Additionally, Mercor—recently awarded a contract with the U.S. federal government—also collaborates with Tencent.

Further correspondence exhibits a similar pattern. Executives from Ant Group, the tech and finance entity behind Alipay, acknowledged a strong partnership with AfterQuery, a U.S. AI training data firm. Conversations with representatives from Alibaba—China’s e-commerce powerhouse—pointed to established contracts with both AfterQuery and Mercor. There’s also evidence that Turing, another data labeling startup from the U.S., has done business with ByteDance, TikTok's parent company.

None of the Chinese companies, including ByteDance, Alibaba, Tencent, or Ant Group, responded to requests for comments. Mercor chose not to provide feedback, but AfterQuery and Surge both maintain they do not disclose customer information. Turing mentioned in a statement: “We are at the forefront of AI development, encompassing both proprietary and open-source models. We anticipate growth in both sectors, with some of today's leading open-source models emerging from non-U.S. labs.”

Collectively, the leading six Chinese AI labs are estimated to spend around $500 million annually on American data labeling services, as reported by two entrepreneurs in the sector citing insights from Tencent and ByteDance executives. Observations from industry insiders indicate that these laboratories see themselves not as direct competitors but as a collective seeking high-quality Western datasets.

American data labeling firms appear willing to sell to this market. Reports suggest that Mercor’s revenue from Chinese labs accounted for about 2% of its total income (with annual revenue surpassing $2 billion by June). At the same time, a significant chunk of AfterQuery's revenue, at least $50 million, arises from Chinese AI companies. Surge AI has been proactive in approaching these clients, with CEO Edwin Chen even visiting China to forge relationships.

Data providers offer bespoke, tailored projects to Chinese AI labs, along with what are termed "off-the-shelf" (OTS) datasets. These pre-packaged training sets are designed for resale to various labs. While this might seem harmless, off-the-shelf datasets are quite powerful; they are developed using robust "knowledge pipelines" created during custom engagements with firms like OpenAI or Anthropic. Consequently, Chinese companies that purchase these datasets gain access to the expertise, rigorous quality checks, and evaluation criteria that have contributed to the training programs of major U.S. AI labs, drastically reducing their developmental timeline.

Cai emphasized the significance of off-the-shelf datasets: “The structure of this data frequently originates from the same vendors who shaped initial scaling strategies for firms like Anthropic,” he noted. “It’s incredibly straightforward for a vendor that has worked with Anthropic to generate similar datasets for a Chinese lab.”

The lucrative nature of off-the-shelf datasets—due to their ability to be resold to multiple clients—generates notable profit margins. Chinese AI labs distinctly express to American data suppliers their desire to procure all data that U.S. labs have already acquired, according to an executive familiar with interactions with these labs.

Max Gazor, founder of Striker VC, which has invested in numerous American AI ventures, acknowledged the ethical complexities surrounding this marketplace. “It's fundamentally an ethical choice regarding whether a company should sell data to Chinese AI outfits,” he remarked.

The regulatory grey area

The U.S. has devoted considerable effort to building barriers around advanced semiconductor technology. However, the vital human expertise required for training AI continues to flow across borders with little oversight.

This situation has bolstered the growth of Silicon Valley’s data companies, many of which are now valued in the billions. Mercor, founded in 2020, is aiming for a valuation near $20 billion, while Surge AI is estimated to exceed $25 billion. Additionally, AfterQuery has seen its revenue soar from $100 million to hundreds of millions in just a few months. Although most of their earnings still come from established U.S. businesses, the appeal of international sales as a lucrative revenue source is undeniable.

There are concerns among some commentators that selling datasets to Chinese firms poses risks to national security. “Some data firms are collaborating with foreign adversaries, and the repercussions are evident,” noted Ali Ansari, CEO of Micro1. “It’s hypocritical to assert American AI leadership while selling millions worth of data to nations we consider adversaries.” 

Micro1 asserts that it refrains from selling data internationally, whereas Scale AI had previously explored a contract with ByteDance but withdrew in 2024 due to security apprehensions.

Conversely, proponents of data sales argue that restricting these transactions could stifle international open-source innovation.

“Effectively limiting data exports or AI transfers could damage American open-source efforts,” Cai pointed out. “Such actions would potentially give an unfair advantage to OpenAI and Anthropic, which comes with its own set of challenges.”

American Startups Are Enhancing China's AI Capabilities

A dim, moody server room bathed in deep reds and charcoal grays, two silhouetted engineers at sleek workstations typing furiously as lines of glowing code arc from their laptops into a central AI core rack emblazoned with a single Chinese flag, thick cables snaking across the floor, shadows and dark backlighting heightening the secretive atmosphere.

These American Startups Are Making China's AI Smarter - Forbes

- U.S. data-labeling companies are increasingly supplying training data to China's AI industry.  
- Chinese firms like Tencent are purchasing sought-after datasets for advanced AI development.  
- The U.S. lacks regulations on data sales to China, impacting the global AI landscape.  

getty

When U.S. sales teams from Silicon Valley’s big data-labeling startups visited this year’s International Conference on Machine Learning in Seoul, Korea, they arrived prepped to court the industry’s big spenders. The data companies—collectively worth tens of billions of dollars and generating billions in annual revenue supplying training data to customers like OpenAI and Anthropic—are used to chasing AI labs that are notoriously demanding, fickle and difficult to satisfy.

They found another eager customer waiting for them: China’s AI industry.

Some Chinese companies have shopping lists. Tencent—which has previously been designated by the U.S. government as associated with the Chinese military, a characterization the company disputes— circulated with prospective vendors a detailed request for training data spanning finance, cybersecurity and one of AI’s most coveted research goals: AI systems capable of improving themselves.

While the U.S. has long prohibited China from buying the chips that power America’s top-tier AI models, it hasn’t taken the same precautions with the data required to train them. This packaged human expertise, produced by specialist networks and shaped by task designs, grading rubrics and quality controls, is part of the expensive data infrastructure that teaches models to handle difficult professional tasks like financial modeling and coding. The business is worth hundreds of millions of dollars annually and it’s helping Chinese AI models close the gap with their American rivals.

“After the compute needed to train the models, good quality data is the most important thing,” Nathan Lambert, a former research scientist at the Allen Institute for AI, told Forbes. “As we try to get models to move to new, more challenging domains, getting good data is the highest-leverage item that can make the model become viable. “

Documentation and communications from buyers at Chinese AI labs and employees from top Silicon Valley data platforms such as Surge AI and Mercor reviewed by Forbes reveal a quiet multi-hundred-million-dollar trade in training datasets. The same U.S. startups serving as suppliers to OpenAI, Anthropic, and, in some cases, U.S. federal entities, are also supplying Beijing’s top labs. Increasingly, Silicon Valley is training both sides of the AI arms race.

For Chinese AI labs chasing American AI performance, there are really only three fast-track moves. The first is to poach researchers. The second is distillation, harvesting outputs from ChatGPT or Claude and using them to train Chinese models. While not illegal, the practice explicitly violates OpenAI’s and Anthropic’s Terms of Service and recently was flagged by the Trump administration’s science and technology advisor Michael Kratsios. A third is easier and comes without potential regulatory headaches: buy the same training data the American AI labs are buying from the same data vendors they are buying it from.

Teaching an AI model to handle complex accounting or web lab protocols isn’t just about collecting the raw work output. The true secret sauce sits inside the data specifications: what type of data to focus on, as well as intricate rubrics written by human professionals that push and prod the model into learning to navigate white-collar work. By purchasing these datasets, the Chinese labs efficiently buy the hard-won, proprietary judgement powering Silicon Valley’s best models.

AI data consultant Sean Cai, who publishes an eponymous data and reinforcement learning blog, puts it bluntly: “If you actually study the shape of the data industry in China and then the entire data supply chain in general, you’ll realize that yes, a lot of the same U.S. data that powers U.S. model advancement is getting sold to Chinese labs as well.”

The $500 million secret data supply chain in China

The pipeline runs straight from the American data labeling companies to China’s biggest tech giants, including firms like Tencent, the Chinese company flagged by the Pentagon. In one message seen by Forbes, a Tencent data procurement executive noted that the company relies on Surge AI–whose customers have included the U.S. Army, the U.S. Air Force and Anthropic– for data provision. Mercor, which late last month told its employees it had won a contract with the U.S. federal government, also works with Tencent.

Other communications reviewed by Forbes show a similar overlap. In text exchanges, an executive from Ant Group, the Chinese tech and finance company behind Alipay, acknowledged a close working relationship with U.S. AI training data outfit AfterQuery. Audio transcripts from conversations with a Chinese ecommerce giant Alibaba employee point to purchase agreements with AfterQuery and Mercor. Internal project documentation suggests US data labeling startup Turing has worked with ByteDance, TikTok’s Chinese parent.

ByteDance, Alibaba, Moonshot, Tencent and Ant Group didn’t respond to requests for comment; Mercor declined to comment. AfterQuery and Surge said they don’t disclose customer information. In a statement, Turing said: “We power the frontier of AI, which includes both proprietary and open source models. We believe both will keep growing, and some of the strongest open source models today come from labs outside the U.S.”

All in all, the top six Chinese AI labs spend about $500 million with American data labeling companies per year, according to two data labeling entrepreneurs who have received market size estimates from Tencent and ByteDance executives. According to one person who has interacted with a variety of them, Chinese AI labs don’t view themselves as being in direct procurement competition with each other; rather they see themselves as a collective buyer pool looking for top-tier Western datasets.

U.S. data labeling companies are happy to sell to them. According to a source familiar with the matter, Mercor’s Q2 revenue from Chinese AI labs was 2% of total revenue (Mercor’s annualized revenue crossed $2 billion in June), and the company has hired a specific engagement manager for China. A larger portion of AfterQuery’s business, at least $50 million in recurring revenue, comes from Chinese AI labs, sources told Forbes. And Surge AI, an early pioneer in the region, has actively courted these buyers, with CEO Edwin Chen traveling to China to meet lab executives directly.

Data labelers sell Chinese AI labs custom, one-off projects as well as what are known as "Off-the-Shelf" (OTS) datasets: standardized, pre-packaged training sets that are meant to be resold to multiple labs. Sounds harmless enough, but off-the-shelf datasets are far from innocuous. They are manufactured with the same "knowledge pipelines" developed during custom engagements for OpenAI or Anthropic. That means the Chinese companies that buy them benefit from the same PhD expert networks, quality control algorithms, and post-training rubrics that help inform the training done at the big U.S. AI labs. And that is enormously helpful. In many cases, Chinese labs can bypass months of trial-and-error by buying OTS datasets with battle-tested reasoning structures.

"You've got to realize the shapes of this data sometimes come from the same companies that tailored their initial scaling directions for Anthropic," Cai, the AI data consultant, explained. "It is extremely easy for a data vendor that has served Anthropic to use the same stack to produce data for a Chinese lab."

This makes off-the-shelf datasets a tempting financial engine for data vendors. Because they can be resold to multiple AI labs, they carry the highest profit margins. And Chinese AI labs explicitly tell U.S. data labelers that they want to buy all the data that the U.S. labs have already purchased, said one executive who has spoken to Chinese AI labs.

Max Gazor, founder of Striker VC, which has invested in many of the American AI model builders, acknowledged this market reality comes with certain complicating factors. “I would distinguish it as more of an ethical decision of whether the company would sell the data to Chinese AI companies,” he said.

Regulatory Wild West

Washington has spent years building a geopolitical fortress around advanced chips. But the human expertise used to train AI manages to cross borders with almost no scrutiny at all.

That’s helped make Silicon Valley’s top data platforms multi-billion-dollar titans. Founded in 2020 and 2023, Mercor is currently eyeing a valuation of $20 billion, while Surge AI is reportedly worth at least $25 billion. Meanwhile, relative newcomer AfterQuery has grown from $100 million to hundreds of millions in annual recurring revenue in three months. While the vast majority of it still comes from established American giants, foreign sales are likely an irresistible secondary revenue stream.

Some say selling datasets to China is a matter of national security. “Some human data companies work with foreign adversaries. and the results show today in Kimi K3,” Ali Ansari, the company’s 25-year-old CEO, recently posted to X. ”We believe it’s shameful to claim American AI dominance desires while selling millions worth of data to countries that we are in adversarial competition with.” Micro1 says it does not sell data outside the U.S. Scale AI also flirted with a ByteDance contract before scrapping it in 2024 over national security concerns.

Defenders of the practice argue clamping down on data sales would choke off global open-source innovation.

“One cannot easily restrict exports of data or AI in a way that does not materially hurt American open source,” said Cai, noting that he does not have a position on the issue. “It gives an unfair duopoly to OpenAI and Anthropic, and that has its own problems.”

These American Startups Are Making China's AI Smarter

In a striking move within the tech community, Jeff Dean, former chief scientist at Google, unveiled the pitch deck for his new artificial intelligence startup, Discovery Loop, on X this past Wednesday. The slides showcase an impressive portfolio of accomplishments that many founders would aspire to emulate.

One notable slide highlights the groundbreaking products developed by Dean and his team, including Google Search—the world’s largest search engine—Google Ads, which represents a significant revenue stream for the company, and Gmail, recognized as the leading email service. They also contributed to Gemini, a series of advanced AI models from Google, and TPUs, the company’s proprietary chips designed for high-performance computing.

Dean outlined the startup's mission to automate the experimental process, emphasizing its applicability across various scientific and engineering domains. While the initial focus will be on machine learning research and engineering, he believes this methodology could address critical challenges across several fields. 

Alongside Dean, the startup is co-founded by other notable figures from Google, including Sanjay Ghemawat, Quoc Le, and Oriol Vinyals, with early investment discussions indicating interest from a select few venture capital firms.

Discovery Loop aims to drive scientific advancements in diverse areas, including computer hardware development and pharmaceuticals. Dean mentioned that initial funding would be co-led by Khosla Ventures and Radical Ventures, with participation from Alphabet and additional investors.

Dean's exit from Google marks the end of a lengthy 27-year tenure and is part of broader changes in AI leadership, which have also seen Demis Hassabis transition from CEO of Google DeepMind to the role of Alphabet’s chief scientist and chair of Google DeepMind.

His departure has sparked a nostalgic response among Google employees, who have taken to internal forums to share fond memories. The broader online audience mirrored this sentiment, with many praising Dean's pitch deck as a monumental achievement. 

Former Google product leader Bilawal Sidhu noted on X, "This will be remembered as one of the most impressive pitch decks ever created." Meanwhile, other industry professionals humorously commented on Dean’s star power in venture capital relationships, signaling that he is the one making the selections rather than the other way around.

A separate slide from Dean's presentation illustrated the founders' collaborative history with key players in the AI landscape, featuring notable names such as Anthropic's Dario Amodei, OpenAI’s Ilya Sutskever and Wojciech Zaremba, and Mistral's Arthur Mensch, among others. 

Another slide showcased the academic prowess of the Discovery Loop founders, highlighting their impressive Google Scholar rankings that position them as leading contributors in the AI research community. Delali Azumah, co-founder of Delalify, succinctly captured the sentiment on X, stating, "Four legends in one room—who wouldn’t want to team up with them?"

Jeff Dean's latest startup presentation slides showcase an incredible display of confidence.

A bright, sunlit startup boardroom with floor-to-ceiling windows. At the front stands Jeff Dean—dark curly hair, glasses, in a sharp suit—confidently gesturing toward a massive slide projected on the wall. The slide explodes with a towering gold arrow breaking through the ceiling, bold metrics like “1,000% ROI” and “🌟 Insane Performance,” and sleek chip diagrams. A semi-circle of executives in silhouette look on, jaws dropped in awe. Warm natural light, minimal tech clutter, epic “flex” atmosphere.

Jeff Dean's New Startup Pitch Deck Slides Are an Insane Flex - Business Insider

- Jeff Dean shared his AI startup pitch deck, showcasing impressive products and achievements.  
- Discovery Loop aims for breakthroughs in AI across computer hardware and drug discovery.  
- Founders include renowned experts, boasting significant academic credentials and industry influence.  

There are pitch decks, then there are Jeff Dean pitch decks.

Google's departing chief scientist shared slides for his new AI startup, Discovery Loop, on X on Wednesday — and they include a list of achievements any founder would be envious of.

One slide lists products the founding team helped build or worked on, including Google Search (the largest search engine), Ads (a highly lucrative business), and Gmail (the most widely used email service).

Other products include Gemini, Google's family of frontier AI models, and TPUs, the tech giant's custom-built chips.

Our general approach is to automate the experimental loop. We think this approach is broadly applicable across many different fields of science and engineering. We’ll initially focus on ML research and engineering, but believe the approach can help with important subproblems in… pic.twitter.com/U5SovZXWrU — Jeff Dean (@JeffDean) August 5, 2026

Dean said the startup, which he is founding with other departing Google heavyweights Sanjay Ghemawat, Quoc Le, and Oriol Vinyals, had shared the pitch deck with "a handful of VC firms."

Related video LinkedIn marketing leader Jessica Jensen said flexibility will help companies attract and retain top talent

The startup is pursuing scientific breakthroughs with AI across various fields, including computer hardware and drug discovery. Dean said that Khosla Ventures and Radical Ventures would co-lead its initial funding round, with Alphabet and others also participating.

'One of the most stacked pitch decks ever made'

Dean is leaving Google after 27 years, as part of a wider AI leadership reshuffle that also saw Demis Hassabis move from CEO of Google DeepMind to become Alphabet's chief scientist and chair of Google DeepMind.

Dean was one of Google's earliest employees and has a legendary status inside and outside the company. Google staff reacted on internal message boards with fond memes, Business Insider's Hugh Langley reported.

The internet responded in a similar fashion to Dean's pitch deck slides.

"Will go down in history as one of the most stacked pitch decks ever made," Bilawal Sidhu, a former Google product leader, wrote in an X post on Wednesday.

"VCs don't select Jeff Dean. Jeff Dean selects VC," Shubham Sharma, a technical lead at YouTube, wrote in a reply to Dean's X post.

"Your deck should have just had one slide that said google me," James Rizkallah, a product data science manager at the financial technology platform Ramp, wrote in a Wednesday X post.

A separate slide shared by Dean is another huge flex.

It says the founders' teams managed or worked with people who went on to lead some of the most important AI companies: Anthropic cofounder Dario Amodei, OpenAI and Safe Superintelligence cofounder Ilya Sutskever, OpenAI cofounder Wojciech Zaremba, Perplexity CEO Aravind Srinivas, and Mistral cofounder Arthur Mensch, among others.

A third slide drives home the Discovery Loop founders' academic credentials, displaying Google Scholar rankings that place Dean, Le, and Vinyals among some of the most highly cited AI researchers.

"4 legends in one room," Delali Azumah, cofounder of software company Delalify, wrote on X on Wednesday of Discovery Loop. "Who would not want to join them."

Repeat founder Ryan Williams secures $10M in seed funding for an AI startup aimed at private credit managers.

More from AI Startups & Investments

OpenAI purchases presentation company NextSlide

Startups Backed by VC Funding Associated with Ongoing Fraud

5 key points to understand about Jeff Dean's new venture

This startup has secured $6M for an AI tutor designed to help children learn independently.