Flybridge

In Gen AI, experimentation is key to success. To build continuously improving systems, evaluations must be a central pillar—after all, you can't improve what you don't measure. Evaluation systems should be tailored to each company and adaptable based on real user interactions. In the latest Flybridge article, we deep dive into the characteristics of a good evaluation system, types of evaluations, and human vs. AI-based model evaluations. We also discuss current areas of research in the space and the strategic role evaluations have for generative AI systems beyond reliability. Massive thanks to Julia Neagu from Quotient AI, Ian Cairns from Freeplay, Yi Zhang from Relari, and Zhao Chen for the insights, feedback, and quotes for the article. As always, thanks to my colleagues Chip Hazard and Jeffrey Bussgang for their contributions. https://lnkd.in/emndRkXt

Hey Pete McDonald you can photo bomb me and Richard DiBona anytime!! Thanks Pardees Safizadeh for the photos and Flybridge for the amazing event!! Jeffrey Bussgang Chip Hazard Cherae Robinson ❤️

I am SO excited to share that OnRamp announced today we have raised $14.2M across a seed and recent Series A funding rounds. Since Paul Holder and I started out four years ago with just some thoughts on a slide deck, we've put our heart and soul into the idea that friction in B2B customer onboarding wasn't something we had to simply accept - it was a challenge we could solve. We've been dedicated to building an incredible team and product to deliver seamless customer onboarding by automating and streamlining the post-sales process. We're working with some incredible customers. And we're just getting started. The funding will be used to enhance our product's leading automation and user friendly capabilities, helping reduce friction from EVERY B2B customer interaction. We couldn't be more excited and look forward to even greater things ahead! https://lnkd.in/e2_FScwd

Honesty is the best policy! I sat down with Julia Maltby, the Principal at Flybridge Capital Partners to have an honest conversation about the hardest parts of working in venture, the MBA experience, and building personal conviction. Episode 016 of “New to Venture” is jam packed with valuable lessons for the next generation of VCs and fantastic stories! Here are some key takeaways from our conversation: 1. The world will test you with difficult times. Take note of the ones who stick by your side and those who make the extra effort to be there for you. 2. Oftentimes, the best performing deals are the most controversial. If everyone thinks it’s a good idea, there may be more competition than you think. 3. For MBA programs, you can prioritize social, career, or academics, but it's difficult to do all three well. Go into the program with intentions and goals! 4. Ultimately, VC is a game of weighing risk. Gather as much information as you possibly can. Make an informed decision. Then be as helpful to that company as you can. 5. There will come a time when you do tons of research and diligence on a vertical, market, or company and arrive at a dead end. Although it’s demoralizing, understand all that work will be helpful down the line. Maybe it’s similar to another sector, maybe it helps you get to “no” faster, or maybe a new law passes and flips the entire market on its head. Write down key learnings from your research so you can refer back to it when needed. Listen on Spotify: https://lnkd.in/eiFJMpke Listen on Apple Podcasts: https://lnkd.in/e6q_zUJ6 #NewtoVenture #VentureCapital #Startups #VC #CEO #Business #Investing #recruiting #networking

🎉 One year ago today - I had no paycheck. 🗓️ One year ago - Mark McQuade Jacob Solawetz and I were figuring out if we should place our bets on small language models (SLMs) and if the market would eventually shift from large language models (LLMs) to domain-specific models, or if closed-source providers would rule on their own 🌙 One year ago - I woke up every night around 3 am, wondering if we would get funded and if I was just chasing a dream that would never pan out. 💡 One year ago - We had a vision of bringing SLMs to the world, but everyone was so excited about Retrieval-Augmented Generation (RAG) that we didn’t know if they would care. 🚀 A year later, we are category-creating the first training solution for SLMs with true differentiation - evo model merging, and spectrum. 💰 A year later, we have millions in revenue and customers from all walks of life: startups, enterprises, financial firms, large P&C insurance companies, software companies, healthcare facilities, automotive manufacturers, and more. 🤝 A year later, we have 30+ "Arcenians" who I am so lucky to call teammates. They believe in our mission and the work we are doing to deliver performant domain-specific models that work with fewer hallucinations and at much lower costs. 🏢 A year later, our company is valued at $120M. 💪 A year later, I still wake up at 3 am... I couldn’t be more excited for the next chapter Arcee.ai and a big thank you for all that have made it happen along the way and for Joseph F. and the entire emergence team for the acceleration on making building Arcee Mary MacCarthy Nora He Michael Nason Shamane Siri, PhD Vladimir Karpukhin Malikeh Ehghaghi Tyler Odenthal Luke Meyers Charles Goddard Traun Leyden Erik Orth Rivindu Weerasekera Hayley Brodeur Nestoras Stefanou Lucas Atkins Chris Smith Maziyar P. Shahrzad Sayehban Anneketh Vij Joseph Maher Dante Simon Fernandes Neto Molly Niland Mike Ryave Michael Del Balso Jeff Boudier David Yehle Yazan "Yaz" El-Baba Gordon Ritter Ben Epstein Alan Chhabra Matthew Bishop, Ph.D.

It is now well understood that application software is evolving from a tool that enhances human productivity to one that autonomously delivers outputs previously generated by humans. However, this evolution’s impact on software pricing — from users or seats to deliverables — is evolving more gradually. In a world where AI-powered applications are delivering tangible outcomes, why do we continue to see pricing models reminiscent of traditional SaaS? 🤔 The short answer - outcomes based pricing is complex and nuanced. More thoughts below.

In 2010, Andres Rodriguez, the co-founder of Nasuni, had a vision for how the cloud could support a distributed, global file system that would consolidate, manage, and protect data at scale, driving significant operational benefits for customers. Alongside Paul Flanagan of Sigma Prime Ventures and Jeff McCarthy of North Bridge Venture Partners, we invested to help bring that vision to life. Today, we are thrilled to see the company’s success recognized by Vista Equity, KKR, and TCV in a transaction that values the company at $1.2B. Importantly, along the journey Paul moved from Sigma Prime to become Nasuni’s CEO. He built a world-class team and together they were manically focused on driving product innovation and customer success. Doing so resulted in top decile customer retention rates, industry leading NPS scores, and a consistent 30% growth rate over many years. The entire Nasuni team has truly built an enduring growth company, for which we are deeply grateful, and we know this history of growth and innovation will continue with the company's new investors! https://lnkd.in/eYPXQVD5