Decibel Partners

You really can't bet against gravity - the funds have to flow! #gravity #funds #flow https://lnkd.in/eYjdrhAN

My first time back to Singapore in ~10 years and it was great to see old friends and make new ones! Hosting our first Sovereign AI Summit in Singapore, we were excited to meet with thought leaders from enterprise tech, major cloud providers, and government in the region. Of course, re-connecting with the local innovators and early adopters at Razer Inc., Grab and Singtel were all highlights for me and our team! Singapore is an innovative thought leader and we look forward to continuing to deepen our engagement and partner with this vibrant ecosystem to support early stage innovation. To that end, we are coming back for SWITCH in October and starting to make plans now - details coming soon… 🔥

New episode of #NeverTooEarly out now. Tips for a successful Cofounder Marriage ❤️ Thank you Nir Polak for sharing your candid wisdom and insights 👫 👬 👭 Substack: https://lnkd.in/efBwRMxP

>115 degree heat in Vegas was appropriate weather for the Black Hat conference this year - could there be anything hotter on the planet than AI + Cyber??? We are grateful to the >600 friends in the Decibel family who attended our panels, meetups, competitions, and cocktail events throughout the week. Cybersecurity challenges can never be tackled in isolation, and there are more ways to win when our offense and defense work together as a team A special thank you to all of our Decibel founders who made this year our biggest and best Black Hat since founding our firm. We look forward to seeing you all soon!

We were excited to announce the winner of our first “Man vs. Machine” competition at Black Hat, an open challenge that gave >100 security researchers the opportunity to compete by hacking AI models in real world simulations. Participants were given 12 different “capture the flag” (CTF) exercises ultimately generating >2 million API requests against widely used LLMs. More than 50 researchers were able to complete the first challenge, while only 3 were able to ultimately complete all 12. Congratulations to Kenneth Yeung who was crowned the winner of the event and was awarded an NVIDIA RTX-4090 by our honorary guest HD Moore (founder of Metasploit and runZero). HD has long been an inspiration for many hackers turned cybersecurity professionals, and we are all hopeful that competitions like these enable researchers in our community to design protective measures and guardrails for today’s widely used AI models. Man vs Machine was a collaboration across the Decibel portfolio, and we would like to thank our friends at Dreadnode (Will Pearce, Nick Landers, Brad Palm), SpecterOps (Jared Atkinson, Will S., Lee Chagolla-Christensen, David McGuire, Jason Frank), AI Village (Sven Cattell) and my partner Alessio Fanelli for co-hosting the event. We look forward to the next competition!

I am excited to be back in Vegas and am looking forward to kicking off Black Hat tomorrow with Mike Privette (Return on Security), Ross Haleliuk (Venture in Security), and Kaiti Delaney (1011 Ventures) at the Innovators and Investors Summit! Is the AI hype real? Is our market overfunded or underfunded? Is it time to take a fresh look at how we build and fund cybersecurity startups? Come help us dispel the myths of what new disruptive ideas are getting funded at our panel. Thank you to our friends Jon Oberheide, Caleb Sima, Michael Sutton, Zane Lackey and the Black Hat advisory board for organizing another great event this year. We are excited to see you all soon! (CC: Dan Nguyen-Huu, Jéssica Leão, Stacey Wueste, Lauren Ipsen, and h/t to Alessio Fanelli for creating this AI + Cyber + Vegas image in Decibel colors)

We are excited to announce the first “Man vs. Machine” competition, our online challenge that gives security researchers the opportunity to demonstrate their skills hacking AI models in real world simulations: * Can you compromise the data in a loan application model? * Can you elude the guardrails on an image generation model? * What enterprise apps can an AI model access with a stolen identity? These scenarios represent the future of AI targeted cyber attacks and enable researchers in our community to design protective measures in today’s widely used AI models. This event is a collaboration between Decibel, Dreadnode, and SpecterOps - we have over 100 elite researchers competing live and are additionally adding new advanced challenges this weekend leading up to our awards ceremony on Tuesday at Black Hat. Our guest of honor will be HD Moore, the founder of Metasploit and runZero who will be presenting the prizes to this year’s winners. Many thanks to our friends Will Pearce, Nick Landers, Brad Palm, Jared Atkinson, Will S., Lee Chagolla-Christensen, David McGuire, and Jason Frank and my partner Alessio Fanelli for creating this event - one of many competitive events we will be hosting in the year to come. If you are looking to join our community and participate in this year's event please use the link below. We look forward to seeing everyone at Black Hat!

✨ In last month's Founder Catalyst edition I had introduced the concept of "Managed Services-as-Software (M-SaS). M-SaS businesses are AI powered services businesses that over time drive from low gross margins, while labor intensive, to high / SaaS-like gross margins, when AI intensive 🤖. This month, I wanted to try to directionally quantify the financial and valuation implications of this shift, specifically focusing on two concepts: 👉 AI-Operating Margin Gain (AI-OMG): The gross margin gap between consulting and software business margin that can be exploited by using AI and automation. 👨💻 👉 AI-Multiple Expansion (AI-ME): The valuation gap between consulting and software businesses that can be exploited by using AI and automation. 💹 More detail on the topic in my latest Founder Catalyst post below. As, always 🙏 for reading! https://shorturl.at/4ofv7

The last 10 days have redefined the AI landscape. OpenAI’s GPT-4o mini slashed costs by 60%, Meta’s Llama 3.1 pushed the open-source frontier, and Mistral's Large 2 model set new efficiency benchmarks. Just a year ago, GPT-4 cost $37.5/1M tokens. Now, GPT-4o mini costs $0.3/1M tokens. This rapid cost reduction is a leap towards intelligence becoming "too cheap to meter." For startups, there has never been a better time to enter the AI race. Explore how this paradigm shift unlocks unprecedented innovation opportunities in my latest blog post (🔗 in comments).