Happy Holidays and welcome to issue #25 of our ongoing newsletter series, From the First Row!
Thanks to Grin Lord and Andrew Wilkinson for sending some startups our way this month. That said, no intros are required - any startup can share their pitch here.
New here? Subscribe!
<aside> 🛠️ Our portfolio companies are hiring! Please let us know if you you or anyone in your network is interested in these opportunities:
<aside> 📚 Thank you to Jess Riegel, CEO of Participant for presenting at Deals & Drinks this December! Thanks to our participants for bringing their energetic and thoughtful engagement to the discussion. Sign up here to be included on the invite list moving forward, and access our Deals & Drinks info page **here.**
</aside>
<aside> 🗣 Our next workshop, Fundraising 101: Standing Out in a Tough Market, will be on Feb 9 from 2 - 3:15pm PT. In addition to the basics of fundraising for early stage startups, we will highlight what we’re seeing in the market and welcome your questions. Register here with other founders and investors!
</aside>
<aside> 🍺 Join Minda at a local happy hour event on January 26!
Alliance of Angels and AoA partner Brex welcome you to join them at a happy hour at Herb and Bitter. Meet other local entrepreneurs and hear from Minda Brusse of First Row Partners. Minda and Conor Bradshaw of Brex will discuss trends in startup financing, angel investing and venture capital, as well as answer any questions you may have.
Free tickets here.
</aside>
<aside> 🔏 Sign Everything – Generative AI effectively eliminating the barrier to creation creates a new problem: human identity on the internet. How do we know if something was truly created by that person instead of someone prompting ChatGPT to write in their style? Fred Wilson argues that cryptographic signatures are the answer. They’re uniquely tied to individuals and provide an immutable record of creation. In a world where AI is playing a growing role in our output, web3 could inadvertently be providing the identity management tools we didn’t realize we deeply needed. I
</aside>
<aside> 🤖 Foundation Models: The Future isn’t happening fast enough – A holistic primer on the meaning of foundation models (for example, OpenAI’s GPT-3) and new opportunities emerging in the realm. Most notable is the dearth of tooling to bridge the gap between model (technical innovation) and application (end user friendly use cases). No matter how powerful models get, they’ll always hit a ceiling without the right tooling around data pipelines, data quality, and MLOps.
</aside>
<aside> 💾 The Evolution of Chips – Chips are not only the bedrock of our digital age, but also a pivotal geopolitical issue. With chip manufacturing being brought onshore, companies opting to design their own chips (Ex. AWS), and chips hitting physical performance limits, the entire process of chip design is being reimagined. But before we make changes, we must deeply understand the present. There’s a lot more to chip design than this article, but it’s the perfect primer for the layman to quickly get smart on chip design.
</aside>
<aside> 💰 How expensive is ChatGPT? – Those of you who’ve worked closer to infrastructure have probably been wondering how expensive these fancy models from Open AI are. While the actual numbers aren’t available, this thread takes you through the process of estimating the compute costs behind ChatGPT. The actual number is valuable, but it’s far more useful to internalize how they reached that figure. As more companies are built off of Generative AI, it’ll be critical to keep this methodology in mind to ensure you target markets that will have positive unit economics. Everything seems super accessible today, but remember we’re in the research phase!
</aside>
Trends:
We’ve framed our funding news to match our evolved pillars: Digital Brain (applied data science) and Human Collaboration (the Digital Brain won’t replace everything/us!). This month we’re exploring a DeepFake startup founded by the creators of South Park, a cloud security and permissions platform, and a new Generative AI driven creative suite.
<aside> 🕸️ Human Collaboration | Deep Voodoo, an AI entertainment startup founded by "South Park" creators Matt Stone and Trey Parker, raised $20m led by Connect Ventures, a partnership between agency CAA and venture firm NEA.
</aside>
We were talking about the technical primitives behind Generative AI just last month, and we’re already seeing huge ethical consequences emerge. Deep Voodoo started in the pandemic by helping creators weave in deep fake content, but who can we truly trust with such tools? While large studios use plenty of cinematic effects (ex. we all know the dragons in Game of Thrones aren’t real), what guardrails do we need to prevent people’s identities being used without their endorsement? While AI-based tools are incredible superchargers for creators, we’re quickly running into the need for techniques to distinguish what’s real versus generated – a lot of today’s activity is still focused on generation, but we’re excited to see what novel methods emerge in identification.