Patrick Spaulding Ryan

This is just incredible. Neal Ghosh's demonstration of how #VentureIQ takes deeply technical and complicated content across a range of industrial use cases and ages/stages through an AI-generated conversational podcast (a la #notebookLM) is really mind-blowing. Check out his post, below, and if you haven't seen VentureIQ, yet, let's get you set up to learn more. You can visit VentureIQ.ai or DM any of us over here to book a demo!

2

Data and observations on recent trends in VC deal terms. Along with more "structure" (participating preferred, multiples on liquidation preference, and cumulative dividends), pre-money valuations have dropped significantly more and not recovered in later stage deals but early stage deals are much closer to 2022 pre-money values & raise sizes. https://lnkd.in/g2kcxznr

5

If you work in tech transfer this is a good read-in on how our diligence tech can assess and evaluate assets with minimal inputs. The best way to learn more is to see it live! Reach out to me directly to learn more!

2

Check out this article from Carta on venture deal terms featuring some quotes from yours truly.

16

2 Comments

On the surface, tech transfer operations (e.g. the process of transferring research and IP into commercialized opportunities) and early-stage startups may seem quite different. But when we at 9point8 Collective dug a little deeper, there was tons of overlap which speaks to the essence of zero-to-one development: 1️⃣ Innovation Focus: both tech transfer and early-stage startups aim to commercialize new technologies or solutions that can address real-world problems or create new opportunities. There is no blueprint to follow. 2️⃣ Intellectual Property: tech transfer ventures usually come with robust IP portfolios, including patents and trade secrets, and early-stage startups often focus on developing and protecting their IP to create a competitive advantage. Leverage IP as an asset is a key part of the strategy. 3️⃣ Market Potential: both tech transfer and early-stage startups target markets with high growth potential. They often aim to disrupt existing industries or create entirely new market segments, so targeting the right markets is essential. 4️⃣ Risk Profile: both types of ventures carry extreme risk and ambiguity due to the uncertainties associated with bringing new technologies or business models to market. With all this overlap, we innovated different versions of our diligence and assessment tool #VentureIQ to focus on tech transfer. Instead of a pitch deck as the raw inputs, we used patent filings instead. We wanted to see if our tech could generate credible insight on how raw technologies can be commercialized. The results have been pretty solid so far. From a single patent filing, we can extract technology readiness levels, target markets, competitors and benchmarks with similar technology, and even sources of public and private funding. From what our design partners have validated for us, these are monumental leaps in the development process -- saving thousands of dollars and weeks of time -- all done with #AI. Know of an interesting #techtransfer use case? Send me a DM so we can chat about it!

11

Great article by Marisa Woutersen at WIPR - World IP Review on Nvidia's rise to be the world's most valuable company. She looks at the relevance of their patent portfolio and compares Nvidia's patent portfolio with other Big Tech companies. The article is behind a paywall but my key takeaways are: - It's very difficult to compare a company like Nvidia that basically has one product (monster graphics cards - yes I know they do a bit more) and tech companies like Samsung who have 100s of products. Inevitably Samsung has way more patents. - AMD is probably their most similar competitor, and Nvidia is ahead here in terms of overall patent filings. - Nvidia is clear playing catch-up with patent filings. They have more pending applications than any other company listed. This suggests there is a realisation that the patent portfolio hasn't kept pace with the company value. Of course, it might also reflect the fact that the suddenly have a lot more cash! - Clearly the value in the portfolio will depend on what and where they invest. The article notes that while AI applications represent core value, where else they invest to diversify beyond AI will be key. The brilliant charts in the article provided by Andrew Samm at Patently. EIP - Better, different.

30

Excited for DCVC portfolio co Alta Resource Technologies to emerge from stealth! The design space of biology is many orders of magnitude larger than inorganic chemistry, and Alta uses it to massively reduce the cost of extracting rare earth metals. https://lnkd.in/g_q5UwRF

69

1 Comment

SB 1047 was a step toward complete AI regulation but failed due to its inherent problems where developers of models are burdened to implement requirements under Section 22603 (b). Overall, I believe the requirements would have created a barrier for companies to enter the AI space, and existing developers of models would have had to comply with high standards, which would come at a cost that not everyone can afford, thereby leading to the stifling of innovation. That being said, it's important that the legislators genuinely see the value technology has brought to the valley while promoting Responsible AI to achieve true AI Regulation. Feel free to share your thoughts—whether you agree or disagree.

26

🔍 Is AI regulation focusing on the wrong risks? 📜 Andreessen Horowitz partner Martin Casado argues that many AI regulations today are based on sci-fi fears rather than real-world, practical risks. Speaking at TechCrunch Disrupt, Casado highlighted concerns over poorly crafted laws like California’s vetoed SB 1047 “AI kill switch” proposal. He believes such regulations confuse developers 🚫💡 and hinder innovation by focusing on hypothetical dangers instead of tangible risks. For Casado, it’s not AI that’s the problem—it’s misaligned policy 🛑. He points out there’s already a robust regulatory framework that can guide AI development thoughtfully without stifling progress. “Why should AI pay for social media’s mistakes?” he asks. What do you think? Are current AI regulations helping or hurting innovation? Let’s discuss below! 👇💬 #AIRegulation #TechPolicy #AI #Innovation #a16z #TechDisruption

2

I recently read an article by Charles Plant here on LinkedIn about the correlation, or lack thereof, between patent ownership and scaleup success. According to Charles [link to Charles’ LI post ]there is no significant correlation between the number of patents and key business metrics such as scaleup ranking, financial velocity, or employee growth rate. My own experience on the subject is mixed.. For clean tech and life sciences firms, patents can help, leading to higher employee growth. However, for unicorns and health tech companies, the absence of patents seems to not affect the outcomes. Even for software tech firms, patents don't significantly impact success. Our stance at RiSC usually follows that if the company plans to pursue patents then we like to ask “Please tell us your strategy.” If it’s just registering IP randomly then we don’t think it adds value to a company. If it’s thoughtful and in support of future defense/offense of IP then we’re interested in learning more. If the company is competing in a litigious market then it’s great to hear that they’ve thought it through in advance. In technology and venture media, particularly coming from governments and economic development agencies, often emphasize patents as a growth strategy, but the data suggests that effective marketing and sales are more crucial for business success in the long run. Founders should focus on understanding their markets and customers rather than solely on intellectual property. For many deep tech founders who are responsible for the core technology innovation and need help navigating the market risks from a business perspective, that’s where we come in nicely at RiSC Capital. #ScaleupSuccess #Deeptech #Innovation #Patents #StartupGrowth

2

A new era of deep tech has emerged. First time funds will raise “unheard of” amounts of capital to fuel next gen deep tech startups - producing outsized, superior returns for LP’s compared to the rest. Prudent investors will act on this data and shift investment strategy as LP’s or risk being left behind savvy wealth managers and CIO’s / FO’s who saw this trend begin 4 years ago.

100

8 Comments

🚀 A Guide to LA’s Tech Ecosystem [PART 1: Accelerators] Every week, I meet 30-50 tech founders, innovators, and scientists who are working on different companies. One question always comes up: 👉 What resources, programs, and funds should I know about in Los Angeles? I tried to find a comprehensive tech ecosystem map of resources but seems like nothing of the sort exists out there. 💡 Over the next few weeks, I’ll share some insights about LA’s tech ecosystem—starting with accelerators that are driving innovation across industries. If you worked with a program that was super valuable and you think it should be included let me know and I will cover it in the next parts. 🌟 Hero House Venture Building Program by SmartGateVC Run by our team at SmartGateVC, we focus on deep tech startups emerging from university labs. We partner with founders through a 16 week program - our main focus areas are AI and brain, AI and physical world link and AI and security. Keep an eye out for the launch of the program in 2025. 🚀 Techstars Los Angeles Accelerator Powered by JP Morgan This mentorship-driven program runs twice a year with a focus on sectors like Aerospace, FinTech, and SaaS. Participants get up to $120,000 in funding and access to an incredible network. 💡 FYI: The Demo Days of the current batch was last week. Matt Kozlov and the team did great work with putting together a nice event where you could connect with LA ecosystem. 🎥 Plug and Play Tech Center Los Angeles Specializing in media, entertainment, fashion, and retail, Plug and Play connects emerging technologies to some of the largest innovation hubs in the world. 🎮 A16Z GAMES SPEEDRUN This 12-week accelerator is built for startups at the intersection of tech and gaming. With a <1% acceptance rate, participants receive $750K in funding and mentorship from industry icons. 🌍 Los Angeles Cleantech Incubator (LACI) If you’re building solutions in clean energy, zero-emissions transportation, or sustainability, this might be a good fit. Startups can secure up to $500K in funding. Had a chance to visit the LACI facility earlier this year and it was a quite an impressive space.

280

The times are a changing ... and so are the deal terms; read more below.

5

Exciting to see startups aiming to use AI to discover chemicals, materials, drugs, etc. Semi companies have taken notice: AMD is investing in and Cadence is acquiring them. Semi/EDA companies have broken out by becoming critical path, whether it’s through critical software, packaging, IP, design kits from foundries, or user communities. Aside from just making it easy to make new materials, these “AI for making X” companies also need to articulate a vision as to how they can establish monopolistic positions in their industries.

144

5 Comments

It seems inevitable that as the underlying foundation models become more powerful, the LLM players will seek to justify the enormous investment that has gone into training their models by moving "up the stack", and evolve from an API or chat interface, to async agents. Regardless of whether the foundation model companies end up directly competing with you in your vertical, I believe as a B2B AI startup, you NEED to go through the thought experiment: how do I future proof my company as the underlying models progress? More here: https://lnkd.in/gkdhmVer

184

18 Comments

LLMOps can be boiled down to three things: 1. Make sure the LLM system does what it is supposed to do: eval 2. Make sure the LLM system doesn't do what it is not supposed to do: red-teaming, guardrails 3. Have continuous visibility into the previous two: observability and analytics into dev/test/prod 2 & 3 can be automated or heavily augmented. 1 is the hardest to automate given every LLM application has its own intent. A design pattern for 1 is the continuous articulation of human intent to an AI evaluator, so the AI can execute them. The future of eval tools augment human evaluators by elicit an important set of high-level intent from human and translate that to low-level assertions. Shreya Shankar did a wonderful job with the EvalGen framework. Who else is working on it?

64

5 Comments

Just finished reading "The Next Giant Leap" by Space Capital - a pioneering VC investing in the emerging Space Economy. The report speaks to how SpaceX’s Starship is more than just the most powerful rocket ever built—it is poised to reshape the global space economy in profound ways. With its immense payload capacity, reusable design, and lower cost per launch, Starship will unlock entirely new industries and transform existing business models. But how should industries and businesses evolve to align with this paradigm shift? 📈 Rethinking Business Strategies & Supply Chains - For businesses, Starship represents an era of mass-production over precision engineering. This means moving away from the focus on mass efficiency toward cost-effective, redundant, and scalable production. - In terms of supply chains, this opens opportunities for new players in manufacturing, materials, and space components. The demand for off-the-shelf parts and non-aerospace suppliers will rise dramatically. 💡 New Use Cases & Opportunities   The opportunities Starship unlocks go beyond launching satellites. - In-space manufacturing of high-value products like pharmaceuticals and advanced materials that cannot be created on Earth. - Mega-constellations of satellites for global broadband, expanding on what Starlink has already begun. - In-orbit assembly and servicing, allowing for larger space structures like telescopes or even habitats to be built in orbit, radically reducing costs. - Lunar infrastructure and resource extraction will become more viable, paving the way for sustainable human presence on the Moon and, eventually, Mars. 📊 What Should Businesses Do Now?   1. Embrace Redundancy & Scale – Instead of focusing on one-off, high-cost solutions, businesses should embrace redundancy and mass production.    2. Simplify Design – With Starship’s payload capacity, there is less pressure to optimize for mass efficiency. Simplified, scalable designs will allow for faster production and iteration cycles, akin to how SpaceX rapidly tests and evolves Starship itself. 3. Develop Strategic Partnerships – Collaborating with forward-thinking space startups, suppliers, and even non-aerospace sectors will be essential to capitalize on the opportunities Starship provides. 🌌 A New Era of Innovation   Starship will democratize access to space, lowering the barriers for businesses to create and scale. The next decade will witness breakthroughs in scientific exploration, energy production, and in-space transportation that we can only begin to imagine today. #SpaceEconomy #Starship #Innovation #SupplyChain #Manufacturing #SpaceTech #Entrepreneurship #FutureOfSpace https://lnkd.in/gNdbwXdq

8

2 Comments

Intellectual Property is a hot button for me, both as a creator and as a former contract negotiator who believes in more than FAR data rights. It’s cool to see how IP gets negotiated when it comes to emerging tech, which is typically behind the power curve with IP policy. Take AI audio, for example. I hear a lot of audio narrators worrying about AI audio taking their jobs, but yesterday, I saw some new business arrangements that accounted for consent and consideration (payment). Some of the complaints I’ve heard in the past assume that if I use AI audio, I’m taking work from them. I’m not. I’m just never going to have someone else narrate a book for me. I’m artsy like that where I want to do my own and have control over my own creation. Maybe add some fun voices or have a full cast of voices. In late 2022, inspired by Dan Ward's self-narrations, I paid for a masterclass in how to create my own audiobooks, bought all the equipment, and even created my own sound room…a week before the upstairs neighbors-from-hell moved in. I’ve since moved but I’ve found other options, including AI voice synthesis/cloning of my own voice. I’m playing around with AI audio, both for the narration and for post-production edits to meet ACX requirements. New territory for me, and fun. Learning a lot. But I had to increase my audio AI tool subscription to put together my first two books (contracting/acquisition-related freebies for y’all). I got to 599,700 out of 600,000 characters and ran out! I didn’t want to wait 3 weeks to renew my monthly subscription, so I decided to do a bunch of audio for the next month and “tiered up” again. The next tier allowed me 2M characters and 600+ voices to choose from. In choosing, I found that some were voice-over narrators and professionals who had cloned their own voices. Some had 30 days to 2 years agreements. (I like that as a creator; not as thrilled as a user.) Some required an extra percentage of characters to use their voice. Others had other arrangements. Whether it’s lucrative for them, I don’t know, but I’m pleased to see different arrangements and opportunities for creators in an AI world. Let’s be creative in our contracts and think beyond FAR data rights. #contracting #procurement #acquisition #datarights #intellectualproperty

6

Is prompt engineering dead? Not yet, but it’s a great example of the speed of AI advancements and the subsequent challenge of navigating tech waves with what can appear to be rapidly decreasing marginal returns. Prompt engineering refers to the process of structuring prompts to increase the accuracy and effectiveness of LLMs. It’s a hot topic at the moment, as it helps solve for many of the quirks that result in poor quality outputs. Today prompt engineering is done by teams of people and is one of the new roles to emerge with this wave of AI. But recent research is showing that LLMs appear capable of creating better outcomes than humans when prompting themselves. Challenge 1: LLM models are being updated / enhanced so frequently that it makes it difficult to build lasting frameworks for one model. Challenge 2: As LLMs get smarter, prompt engineering capabilities will be embedded into the models themselves and become a feature of the LLM. Does this make investing in prompt-engineering today a waste - heck no. It just needs to be part of a larger LLMOps and ultimately DataOps motion. Why, LLMs don’t speak English, they just do a lot of math. And ultimately having *usable* proprietary data and a killer overall platform is the key to winning over the long run. We’ve been chatting to a lot of founders lately on their strategies for navigating many of the challenging questions when embedding AI into their platforms: should I fine-tune my own model? Which model should I use? Should I build a prompt-engineering team? We definitely do not have all the answers but thankfully we know a lot of folks who are actively navigating these questions and / or building the scaffolding to help others do so. We’re working on a short essay on some tips and tricks for early & growth stage companies - if you’re interested in contributing please reach out! #ai #data #software

28

1 Comment