Mayfield

From a spark of an idea to breakthrough AI solutions — we're delighted to unveil the winning teams of the inaugural Mayfield AI Garage at Berkeley! After reviewing hundreds of applications from brilliant student builders and researchers across the UC Berkeley community, we selected five exceptional teams pushing the boundaries of what's possible in AI. Each team will receive a stipend, mentorship from Mayfield and our partners, and compute credits to bring their visions to life. Congratulations to our 2025 winning cohort: - Touchstone – Evelyn T., Emily Tran An AI teammate to prevent refund fraud in e-commerce - Interlinked – Rehaan Bicha, Rushil Desai, Neel Shetty An AI teammate advancing fire and environmental safety - Oratora – Trishala Jain, Mark Nour An AI-powered coach for interview preparation - NovaSky – Dacheng Li, Shiyi Cao, Shu (Lynn) Liu, Tyler Griggs, Simon Mo Open-source, state-of-the-art AI models that are affordable and accessible - TAG (Table-Augmented Generation) – Asim Biswal, Liana Patel, Shu (Lynn) Liu, Sid Jha, Amog Kamsetty A new approach to answering natural language questions using structured data Huge thank you to our partners at UC Berkeley’s CDSS, Berkeley SkyDeck, Microsoft, and NVIDIA — your support made this possible. Learn more about our winning teams and other Mayfield AI Garage initiatives: mayfield.ai/garage Berkeley SkyDeck, Darren Cooke, Caroline Winnett, Sibyl Chen, Chon Tang, Brian Bordley, Sanjay Velappan, UC Berkeley College of Computing, Data Science, and Society, Jennifer Chayes, Tiffany Lohwater, Rebecca Ricksen, NVIDIA for Startups, Howard Wright, Sydney Sykes, Vanessa Ching, Natasha Pollayil, Microsoft for Startups, Pamela Fox, Rob Ferguson, Marcelo Bossi, Courtney Sales, Kelvin Keane  #AI #Berkeley #AITeammates #FoundersFirst

What started as an idea last year and launched in January has now culminated in something special — we’re proud to announce the winning teams of the first Mayfield AI Garage at Berkeley! After receiving hundreds of applications from student builders and researchers across campus, we’ve selected five teams to receive a stipend, mentorship from our team and partners, and compute credits to build their groundbreaking AI Ideas. Congratulations to: - Touchstone – Evelyn T., Emily T. An AI Teammate for e-commerce refund fraud prevention - Interlinked – Rehaan Bicha, Rushil Desai, Neel Shetty An AI Teammate for enhanced fire and environmental safety - Oratora – Trishala Jain, Mark Nour An AI Teammate for interview preparation and coaching - NovaSky – Dacheng Li, Shiyi Cao, Shu (Lynn) Liu, Tyler Griggs, Simon Mo Open-source, state-of-the-art AI models that are affordable and accessible - TAG (Table-Augmented Generation) – Asim Biswal, Liana Patel, Shu (Lynn) Liu, Sid Jha, Amog Kamsetty A new paradigm for answering natural language questions over structured data We’re so proud of this first cohort — and grateful to our partners at UC Berkeley’s CDSS, Berkeley SkyDeck, Microsoft for Startups, and NVIDIA for helping bring this program to life. This is just the beginning. These teams embody the future of AI: ambitious, responsible, and human-first. Watch the video of our celebration with the winners below. Learn more about the winners and the Mayfield  AI Garage here: https://lnkd.in/gAQCUPc8  #AI #Berkeley #AITeammates #FoundersFirst

What happens when two former Tesla engineers decide to fix one of the most painful parts of content creation? You get Mosaic — and they just won the Mayfield AI Garage competition at SaaStr. Here's what caught our attention: Co-founders Adish Jain and Kyle Wade didn't just build another video editing tool. They created an AI platform with a "canvas" interface that lets you build multimodal video editing agents. Out of 200+ applications and 50+ live demos at the Mayfield AI Demo Stage, Mosaic stood out as an audience favorite because they're solving a real problem that every creator faces. Video editing is still heavily manual and painfully tedious. Huge thanks to Jason Lemkin, Amelia Lerutte, and the entire SaaStr team for partnering with us and bringing the Mayfield AI Garage to SaaStr. Adish Jain, Kyle W., Jason M. Lemkin, Amelia Lerutte, Navin Chaddha, Mosaic #SaaStr #AI #Innovation #Startups

Congratulations to our Managing Partner, Navin Chaddha, who made the Forbes Midas List for the 17th time! This recognition is a testament to his people-first leadership that drives everything we do at Mayfield. Forbes recognized Navin among a handful of VCs with “remarkable staying power,” highlighting those with uninterrupted Midas streaks spanning over a decade. This consistency reflects our unwavering commitment to empowering founders who are building the future. Navin’s approach has always been: invest in people first. Here’s to another year of partnering with founders who are changing the world. 🚀 #MidasList #VentureCapital #PeopleFirst

What if solving math problems is the path to truly intelligent machines? Tengyu Ma’s journey comes full circle from winning a Silver Medal at the International Math Olympiad to pioneering mathematical reasoning in large language models at Stanford. I had the pleasure of hosting Tengyu on the latest episode of #AIPathfinders. Tengyu is a rare triple threat: Assistant Professor of Computer Science at Stanford, Chief Scientist at MongoDB, Co-founder of Voyage AI (recently acquired by MongoDB) We covered a wide range of topics, including: ▪️ Voyage AI’s origin story – why he built a company around embeddings and re-rankers, the backbone of AI-native search and RAG systems. ▪️ RAG vs. fine-tuning vs. long context – why modularity and cost-efficiency make retrieval-based methods a winning bet today. ▪️ From lab to market – how Tengyu’s lab is developing RL methods that enable models to not just solve problems, but propose and prove new conjectures—pushing LLMs closer to AGI-like reasoning. ▪️ Founder insights from a deep tech perspective • “Start with your edge—core tech. The market will find you.” • “Don’t get hung up on the business model. In fast-moving tech, adaptability is the advantage.” • “You don’t need to wait until you’re ‘ready’—just start. Learn by doing.” ▪️ He also draws compelling parallels between research and startups: •Fundraising = Grant writing •Product marketing = Publishing papers •Sales = Getting others to adopt your ideas And a reminder that stuck with me: “Even without new breakthroughs, we have years of headroom just applying what we’ve already built.” 🎧 Links in comments

“It’s the best time in the world to be a founder in this new AI world.” That’s how John Chambers, the legendary former CEO of Cisco, current CEO of JC2 Ventures, and one of tech’s most respected voices, captured the moment we are in during our Humans in AI conversation. In our latest episode, we dove into why AI isn’t just the next shift — it’s the fastest, most consequential wave we’ve ever seen. According to John, the AI era will move 5X faster than the internet revolution and deliver 3X the impact. Here’s what stood out from our conversation: - 2025 will be a turning point: John believes we’re about to see a shift from experimentation to enterprise-wide implementation. By year-end, AI will be embedded across every function — sales, product, legal, engineering. The companies that move fast will create massive productivity gains. Those that don’t? They risk becoming irrelevant. - Sovereign AI is real and global: We talked about how countries are no longer waiting on big tech. They’re racing to define their own AI clouds, policies, and platforms. From the U.S. and India to China and the EU, governments understand that leading in AI will define economic strength and national competitiveness. - Founders must build with AI at the foundation: John now invests only in companies that are AI-native. It’s no longer a feature, it’s the architecture. Whether you're building in cybersecurity, healthcare, or enterprise software, the winning teams will treat AI not as an add-on, but as core infrastructure. - Speed is a differentiator and a risk: Founders often say they’re moving fast, but John warned that too often, the biggest bottlenecks are CEOs themselves. His advice? Move with urgency, but anchor it in a repeatable, scalable playbook. Vision without execution is just ambition. - Dream bigger than feels comfortable: If your idea feels too ambitious, it might be a sign you’re aiming just right - “There’s no room in the world for small dreams.” The AI era will reward those bold enough to reimagine how the world works, not just iterate on what already exists. This was one of the most energizing conversations I’ve had on Humans in AI. Watch the full episode with John below and take notes. He’s seen the cycles before, but he’s betting this one will be the most transformational. What’s one bold move you think founders must make to stay relevant in the AI era? #AI #Founders #SovereignAI #Leadership 

Forget "using" AI tools – in the near future, we'll work alongside AI Teammates built on agentic technologies. This isn't hyperbole: search interest for "AI agent" has quadrupled since September 2024, reflecting a fundamental shift in how we'll work. What's new isn't just incremental improvement, but a sudden leap in complexity and capabilities. We're witnessing the birth of systems that can reason, learn, and execute complex tasks with minimal human oversight.  The winners of the next decade won't just adopt AI – they'll fundamentally redesign their organizations around human-AI collaboration. The most successful companies will create AI agents that: 1. Amplify human expertise (not replace it) 2. Navigate physical environments (moving beyond screens) 3. Coordinate in complex multi-agent systems (the new cloud architecture) 4. Build trust through transparent governance (embedded from day one) 5. Transform humans who partner with AI into something more: hum(AI)n The biggest opportunity isn't automating away jobs – it's creating AI Teammates that amplify human capabilities in a true Collaborative Intelligence Era. In my latest newsletter, I break down five critical insights founders and product builders need to understand the agentic frontier, identify opportunities for intelligence and integration, and build with human amplification (not replacement). If you're building an AI Teammate, I'd love to hear about your experience in the comments. #AITeammates #AIagents #CollaborativeIntelligence World Economic Forum

“The biggest mistake in AI right now? We’re not dreaming big enough.”   From GPT Origins to the Future of AI Agents, I sit down with Karthik Narasimhan , co-author of the original #GPT paper, Associate professor at Princeton known for pioneering work at the intersection of language and decision-making, including SWE_bench, ReAct, Tree_of_Thoughts, and TauBench.   We dive into ▫️ GPT's origin story - How a rejected paper at OpenAI sparked a revolution ▫️ Early agent research - from playing Civilization to WebShop ▫️ Unified path to connecting language and decision-making ▫️ How AI agents and human agents will coexist, communicate, and collaborate ▫️ Why agent reliability is one of the big challenges in real-world deployment. ▫️ Practical advice for PhDs and builders commercializing cutting-edge ideas   This one’s for founders, builders, and researchers who want to see past the noise. 🎧 Links to Podcast in comments

While AI models are evolving at lightning speed, the chips that power them remain stuck in a slow, costly development cycle. Cognichip is here to change that with a bold vision to reduce chip development times by up to 50%. Founded by semiconductor veteran Faraj Aalaei and supported by a world-class team of AI experts from Stanford, Google, and MIT, Cognichip is tackling one of the most critical bottlenecks in tech: accelerating chip design through AI. This isn’t just a better EDA tool—it’s an entirely new foundational AI model built to speed up, improve efficiency, and make the trillion-dollar chip industry more accessible. We’re proud to have co-led Cognichip’s $33M seed round alongside Lux Capital, FPV Ventures, and Candou Ventures, to power the future of semiconductor design. Congratulations to the Cognichip team! 🔗 Read more via TechCrunch: https://lnkd.in/g8wRG6AE Faraj Aalaei, Cognichip Inc., Ehsan Kamalinejad, Simon Sabato

Last week at TiEcon 2025, twenty female founders took the stage to pitch their startups in our Mayfield AI Garage "She Pitches, She Wins" competition to empower the next generation of female founders who will build the future of AI. atronous.ai AI, led by Jayashree Dutta, captured both our judges' imagination and the top prize:   • $50,000 stipend  • Up to $350,000 in Microsoft Azure compute credits  • Access to NVIDIA’s Inception program  • Dedicated office space from TiE Big thanks to our expert judges, Patrick Salyer, Sydney Sykes, Manish Chandra, and Steve Herrod, who had the difficult task of selecting just one winner. Congratulations to Jayashree Dutta and the entire Atronous AI team! We're excited to witness your journey and the impact you'll make. Hema Shukla, Swathi Adiseshu, TiE Silicon Valley, Microsoft for Startups, NVIDIA for Startups Video: Imani Sires #Innovation #WomenInTech #FemaleFounders