John Furrier

Big M&A news in Seattle startup land as Palo Alto Networks plans to acquire 3-year-old cybersecurity company Protect AI, ranked No. 10 on the GeekWire 200, our list of top tech startups across the Pacific Northwest. It's part of a growing cohort of cybersecurity startups that are based in the Seattle region or have a significant presence in the area. Read more: https://lnkd.in/gfx_pXkS

91

On the heels of a strong Q4 2025 earnings report, DigitalOcean CEO Paddy Srinivasan says the company has crossed $1B in ARR. It is also accelerating toward 30%+ growth by positioning itself as the “AI inference cloud,” capturing surging demand from startups and enterprises that need full-stack infrastructure to run AI workloads. Full Fortt Knox conversation in the comments:

227

7 Comments

AI and Tokenization unlocks new business models. At Decasonic, we have been seeking investment opportunities at the intersection of AI x RWA. Our investment thesis: Exponential technologies like AI and Web3 converge to build entirely new markets. These new use cases bring the digital world to the Real World Assets (RWA), enabling programmability, composability and liquidity that are native to digital code. The Dune x RWA.xyz RWA Report 2025 (attached) highlights how tokenized U.S. Treasuries alone have reached $7.3B in market size, an 85% YoY growth just this year. This is just one signal of how quickly tokenization is scaling across treasuries, credit, equities and other assets. We believe AI will accelerate this adoption and unlock entirely new business models. Here are three areas where we see Web3 tokens and AI mutually flourish: 1️⃣ New Business Models for AI Creations, AI Agents and AI Digital Twins AI agents, digital twins, and generative models are producing novel assets daily. Tokenization provides identity, ownership, and monetization rails, allowing creators and enterprises to build sustainable business models around AI native output. 2️⃣ Proof of Work and Proof of Personhood In an age of infinite AI content, trust becomes scarce. Tokenized attestations validate that work was produced by a verified human or an authenticated AI agent. This secures reputation and enables trust driven marketplaces. 3️⃣ Data Provenance for AI Models are only as good as the real and synthetic data they ingest. Tokenization provides a cryptographic ledger for data lineage, ensuring verifiable provenance, aligned incentives, and compliance for enterprises and regulators. The big idea: AI creates exponential output while tokenization ensures that output is verifiable, ownable, and monetizable. Together, they enable an entirely new layer of the global digital economy. The convergence of AI and tokenization is not just infrastructure. It is the foundation for the next generation of business models, from AI native enterprises to institutional grade financial systems.

18

2 Comments

Amazon's AGI Lab unveils Nova Act, a general purpose AI agent to control a web browser and independently perform some simple actions, and a Nova Act SDK.

20

1 Comment

Today we Lightspeed are excited to announce the industry-first formal partnership with 𝐒𝐤𝐲 𝐂𝐨𝐦𝐩𝐮𝐭𝐢𝐧𝐠 𝐋𝐚𝐛 to support the next generation of entrepreneurs & technical leaders at University of California, Berkeley. Many of the most consequential AI-native startups of our generation were founded by students, postdocs, faculty, and alums of UC Berkeley, including OpenAI, Databricks, Perplexity, Skild AI, ReflectionAI, Physical Intelligence, and more. Sky Computing Lab has followed in the footsteps of predecessors Berkeley RISE Lab & AMPLab to push technical research & commercialization across both AI/ML infrastructure and applications. Generational startups often begin within the walls of academic institutions like UC Berkeley, but the research funding landscape (in particular tied to NSF) is shifting, and the impact is already visible: fewer grants, slower timelines, and heightened urgency to commercialize. On behalf of Lightspeed, James Alcorn & I are honored to partner with Ion Stoica, Joseph Gonzalez, Dawn Song, and the Sky leadership team with a commitment of capital, network, and startup resources to ensure Berkeley's next wave of entrepreneurs has the platform and partners to build industry-defining companies. UC Berkeley College of Engineering UC Berkeley College of Computing, Data Science, and Society UC Berkeley Electrical Engineering & Computer Sciences (EECS) #UCBSkyComputing #BerkeleySky #UCBerkeley https://lnkd.in/etDTiBzC

261

19 Comments

With the Google TV layoffs, some are speculating Google is deemphasizing, or may cut, the TV OS. However, the declining cable TV market and Gemini are more likely the reason for the RIF. https://lnkd.in/gXQfDN3U

35

7 Comments

#TheTenYearFixedTerm. It drives the entire startup ecosystem. Here’s how. Week 4 of “Venture Capital” at the Santa Clara University School of Law. LPs don’t want to be trapped in an illiquid venture capital fund forever, paying annual management fees. How can they get out? There is a secondary market for those interests, but it’s primarily opportunistic, and targets desperate sellers. So without a structural endpoint, they could be trapped forever. And there is: the 10 year fixed term. On that anniversary, subject to possible but limited extensions, the fund must terminate and wind-up, distributing its final assets to its investors. Seems benign, but the impact is massive! 1️⃣ Our #VC and #startup culture is defined by SPEED. Startup-to-growth-to-liquidity, as quickly as possible. Market dynamics are most certainly relevant, and prevalent. But there’s a structural driver as well: yes, the 10 year fixed term. If you are going to fully harvest your fund in 10 years, you need to front-load your investment activity (to ensure that your portfolio companies have at least some time to grow and ripen). So VCs race to deploy their capital within the first 5 years. But that’s not all - they then apply pressure on founders to drive toward exit. Which in virtually all cases, is M&A (sale of the company). The need for LP liquidity within 10 years drives the need for startup speed. 2️⃣ #AgencyCosts refer to the inevitable conflicts-of-interest that arise when one party (the agent) is tasked with acting on behalf of another party (the principal). In the #VC context, the ultimate principal is the #LimitedPartner, and their agent is the #VentureCapitalist. Limited partners want big returns, and liquidity. But what’s to stop the VC from diverting focus (and capital) toward an unending lifestyle instead? In other words, people have their own agendas and motivations. Always. So how does this principal, here, ensure that incentives are aligned and agency costs minimized? Yep. The 10 year fixed term. By dramatically limiting the space for self-dealing by creating a brief window of active existence, the VC is forced to focus on getting to liquidity. They simply do not have the opportunity to selfishly create an endless, self-perpetuating empire underwritten by the limited partners. Bracketed with their carried interest, the VCs are incented to focus on big returns in a concentrated period of time. Exactly what the LPs want! I was on the road this week, and so I recorded the lecture for the students. I normally can’t provide access to my lecture recordings, because they live behind university firewalls, but this week I bootlegged it and uploaded them to my YouTube channel. If you are interested, this lecture (broken into 3 segments) is available there for free - no agency costs in sight! #DWProf

19

10 Comments

Thrilled to see the fruition of almost a year's work in developing the Agentic Real Time Framework (ARTF). This is one of the most forward-looking & transformative technical developments for the digital media ecosystem since... well... OpenRTB, I suppose. ARTF creates a true control plane that supports a future in which autonomous agents collaborate & optimize the bidstream in real time while ensuring rigor, safety & interoperability. ARTF is built on a containerized architecture that supports not only autonomous agents but also specialized software services that enhance the buying & selling of media. Paired with the new OpenRTB Patch protocol, these containerized components can enrich the bidstream with new intelligence, validate it for quality, or propose targeted adjustments with clear, auditable intent. Instead of transmitting full bid requests to external systems, platforms can now orchestrate purpose-built agents & services that apply atomic, transparent changes within tightly controlled environments. This unlocks dynamic deal curation, cohort activation, fraud evaluation, metadata enhancement & entirely new forms of auction-time decisioning, all while preserving privacy, maintaining operational control & enabling far greater extensibility across the programmatic ecosystem with less investment. The broader implications for OpenRTB are compelling. ARTF modernizes real-time bidding for an AI-driven world, dramatically improving efficiency. Programmatic processing costs are reduced. Sustainability improves. Barriers to entry for startups & innovators drop significantly because they no longer need to build & maintain heavy infrastructure to participate. What makes this even more impactful is ARTF's support of the Model Context Protocol. MCP enables model-to-agent communication & future autonomic agent behavior, making this specification foundational to an agentic marketplace where innumerable agents operate together to shape & elevate the media supply chain. ARTF is not theoretical. It is in production at Index Exchange & Chalice AI, whose real-world deployments informed the design, architecture & performance expectations in this specification. We have opened an extended public comment period & welcome industry feedback. Links to the specification, sample code & press release are below. Also, good thing we did not name this the Framework for Agentic Real-Time. The acronym would have been… less than ideal. None of this would be possible without the hard work of Tech Lab members Index Exchange, Chalice AI, The Trade Desk, Amazon, Attain, Basis, Epsilon, HUMAN, Magnite, Netflix, OpenX, Optable, Paramount, PubMatic, TRUSTX, WPP Media & Yahoo. Thank you. Spec: https://lnkd.in/ep3q_CBg Sample Code: https://lnkd.in/esC7M9bd PR: https://lnkd.in/ezfAhFwv

142

10 Comments

Google DeepMind has launched Gemini 2.5 Deep Think, now available to Ultra subscribers in the Gemini app. Official announcement and more details: https://lnkd.in/egvkTt_z This is a good opportunity to better understand what do we mean by "thinking models" and why do we need them. 🤓 💡 If you have 15 minutes to spare, I strongly recommend this excellent short talk from Jack Rae (Google DeepMind's thinking tech lead): https://lnkd.in/eKjKxQVG tl;dr; *AI intelligence progress is all about removing bottlenecks* Historically, major advances in AI come from identifying core “bottlenecks” in current models, and then innovating to remove them (e.g., deep learning, attention, transformers). *Bottleneck today: test-time compute* While LLMs like Gemini and ChatGPT are powerful, they are constrained by a fixed amount of computation per user request (“test-time compute”). This limits the depth and quality of reasoning and solution generation for complex queries. *Solution: dynamic, iterative ‘thinking’ stage* Hence the “thinking stage,” where the model can iteratively loop, perform extra reasoning, and dynamically decide how much additional compute to spend based on the problem’s difficulty. This allows the model to give simple answers quickly and use “deep think” mode to spend much more time and compute on hard or valuable tasks. *Implementation: reinforcement learning for thinking* Models are trained to use this thinking stage via reinforcement learning—rewarding successful problem-solving even with vague, high-level feedback. This shapes the model’s ability to reason, self-correct, decompose problems, try alternative approaches, and use tools when needed. This generic method for training "how to think" led to emergent behaviors such as forming hypotheses, testing them, discarding wrong solutions, breaking down complex tasks, and combining different solution strategies. *Ongoing challenges: efficiency and overthinking* Research continues on making the thinking stage more efficient (avoiding overthinking, unnecessary computation) so that models automatically and adaptively choose the right level of effort for a given task, as well as data efficiency - LLMs that think incredibly deeply and data-efficiently, from a small set of input knowledge. Happy thoughts! #AI #Gemini #DeepThink #GoogleDeepMind #ArtificialIntelligence Antonio Gulli Vladimir Vuskovic

16

Interesting read from Mitchell Troyanovsky from Basis, a Factorial portfolio company: https://lnkd.in/gFWv9iJj TLDR: The PC era created IT departments. The internet era created data teams. The AGI era demands an internal agent team. My main takeaway is that right now companies are thinking more about trying to get employees interact with agents than giving the agents the *context* to interact with employees. If a company's knowledgebase is going to be the equivalent of its codebase, we might need evals, tests, etc to make sure that it's giving the right context to employees to help them do their jobs better, in the same way that we have unit tests on code to make sure we don't introduce bugs. This is harder than you'd expect but anything less will end up with useless output at best and fail states at worst.

17

Big milestone for Doppler Finance: Doppler just raised a $3M seed round led by Reforge, with participation from DCG, Maven 11, HashKey Capital, GSR, CMCC Global and others. Their goal? Unlock institutional-grade yield infrastructure on the XRP Ledger, something that’s been missing since day one. Here’s the opportunity: -> XRP’s ~$200B market cap has less than 0.1% in DeFi. -> Ethereum, by comparison, has ~20% of its market cap active on-chain. That’s hundreds of billions in idle XRP and Doppler is building the rails to activate it. With custody integrations (Fireblocks, Copper, Bitgo), verifiable proof-of-reserves, and plug-and-play APIs for wallets/custodians, they’re building for regulated capital but with open access. Their $500M+ capital pipeline speaks volumes. This is a major unlock for #XRPFi and one of the most promising attempts to bridge compliant yield infra with an underserved but globally significant asset. Excited to see how this one plays out.

96

5 Comments