Alexey Aylarov

For the last 10 years, I’ve been told you can’t build “big enough” companies in deep tech. Too hard. Too slow. Too capital intensive. That kind of value creation, they said, belonged to consumer internet, enterprise SaaS, fintech, or whatever was trending that year. And yet—today, NVIDIA just became the world’s first $4 trillion company. A semiconductor company. Not a crypto protocol. Not a neobank. Not a social app. Nvidia is the prototypical OG Silicon Valley company—headquartered in Santa Clara, not Market Street or Mission Bay or (welp) Miami!. The kind of place where startups were built when I was in the trenches in the ’90s and 2000s. Back when deep tech was the default. Their rise is a salute to that era—of hard engineering, real science, and foundational innovation. And it’s not just Nvidia. Look at the some of the most valuable companies created in the last 10 to 15 years - 🔹 SpaceX – Rewriting the rules of space access 🔹 Starlink – Expanding global infrastructure from orbit 🔹 Palantir – Operationalizing data at national scale 🔹 Anduril – Building the future of defense autonomy 🔹 OpenAI / Anthropic – Redefining what machines can do These are companies solving humanity’s hardest problems—with high stakes, long arcs, and massive outcomes. At MFV Partners, we are proud that we never wavered from this path. We back early-stage deep tech founders not because it’s easy or fashionable (like it is becoming today), but because it’s where the next era of value will be built. Congratulations to Jensen Huang and the Nvidia team. You didn’t just reach $4T—you proved a generation of investors wrong. #DeepTech #VentureCapital #Founders #Engineering #Semiconductors #BuildingTheFuture https://lnkd.in/guB4VuWd

140

11 Comments

The LLM era is peaking. The SLM era is just getting started. Why care? 🐧 Tencent just dropped their 0.5B-parameter Hunyuan Small Language Model that can run offline on a smartphone or edge/constrained device — no cloud, no Wi-Fi, no “please hold while your AI is thinking.” 📈 It’s tiny compared to models like o4 (~200B) or o4-mini (~20B), yet it supports a 256K context window and both “fast” and “slow” thinking modes. Translation: it can produce high-quality outputs while living entirely on-device. Our AIOT portfolio companies like Atym, Mimiq and Tripolar Industries (stealth) are primed to take advantage of this wave. 🏁 Why SLMs > LLMs in certain domains: - Speed: Millisecond responses. - Offline operation: Works in connectivity deserts. - Privacy: No data leaves the device. - Focus: Perfect for specialized tasks. 🤖 Industry will be a killer app for SLMs: - Edge AI in factories: Local analysis of production data without risking IP leaks. - Aerospace & automotive: On-device AI guidance for additive manufacturing. - Frontline productivity: Real-time troubleshooting without a network tether. 🔥 Hot take: Within 3 years, most “AI in manufacturing” will not be powered by giant LLMs in the cloud — it’ll be nimble SLMs at the edge. The next AI arms race isn’t about who has the biggest model. It’s about who can make the smartest model that fits in your pocket. What’s your bet?

192

29 Comments

AI is no longer just a breakthrough, it’s an ecosystem in motion. At my recent visit to Israel, Ben Haklai, National Technology Officer at Microsoft gave us a powerful lens into how AI is evolving from discrete tools to intelligent agents, capable of deep reasoning, self-execution, and context-aware decision-making. Here are a few takeaways that I would like to share: 1. AI Models Are Crossing a New Threshold We saw OpenAI's o1 preview outperform humans in complex math and coding tasks. But this isn’t just about higher accuracy, it's about moving from automation to cognition. The age of AI agents that can reason, not just retrieve, is here. 2. Cloud + Local = Hybrid Intelligence The rise of Phi models and small language models (SLMs) running on mobile, edge, and local devices means AI will increasingly live within your workflow, not outside it. The cloud remains the backbone, but edge is becoming the brainstem. 3. The Jevons Paradox Is Real in AI Too As Satya Nadella aptly said, the more efficient AI becomes, the more we'll use it. Accessibility drives adoption. Commoditization drives creativity. This is both a gift and a governance challenge. 4. AI Is Becoming a Platform, Not Just a Product From Hugging Face to Azure OpenAI, NVIDIA, Meta, and Databricks- what stood out was the vast model marketplace available for developers. The future isn’t one AI, it’s multi-model orchestration, based on context, compliance, and cost. 5. It's Time to Shift from Tech Curiosity to Business Readiness Ben’s session emphasized that agents built in Copilot Studio are being trained not just on language but on your company’s logic. This demands a rethink of everything: roles, workflows, risks, and ROI. The takeaway? We don’t need to ask if AI is coming. We need to ask: Are our teams, tools, and thinking aligned for what’s already arrived? Kudos to the Ben for not showing us flashy demos but showing us the future with humility and clarity. 📍India’s tech ecosystem- startups, enterprises, and policy thinkers must take notes. The convergence of models, cloud, governance, and domain-specific reasoning is not a distant vision. It’s happening now. Let’s not just use AI. Let’s build responsibly, scale wisely, and lead boldly. Thank you Maya Sherman and Embassy of Israel for this wonderful opportunity to learn from the man himself. #MicrosoftIsrael #AIagents #OpenAI #CopilotStudio #GenAI #CloudInfrastructure #DigitalLeadership #AIReadiness #IndiaIsrael #AIethics #PhiModels #TechnologyStrategy #FutureOfWork

181

24 Comments

GPT-5 dropped last week. It’s a solid step up from GPT-4 - especially in agentic workflows and integrated tool use - but perception is governed by the cruel math of expectations minus reality. Sam Altman promised the moon; OpenAI delivered the stars. People noticed the gap. The live launch event did not help. There were chart crimes. Elon live-tweeted Grok 4 beating GPT-5 on select benchmarks. A demo mangled the Bernoulli Effect which isn't something you want to get wrong if you claim to have built a PhD-level expert in everything. The over-promise has so dominated the conversation that it’s obscured what GPT-5 actually is. So let’s set aside the comms postmortem and look at the product: (1)Tool calling as a first-class citizen. Before, GPT-4 could use tools in the way that I can use a blender: I know how to turn it on when I need it, but it’s not part of my thought process. GPT-5 has tools baked into its reasoning - it can call multiple in parallel, merge outputs, and keep context over hours-long jobs. Cursor, Vercel, and Notion all say it now finishes workflows that GPT-4 would abandon halfway through. If you’re building production agents, this is not a small thing. (2) The end of ModelPalooza. No more picking between GPT-4o, o3, and friends. ChatGPT now routes queries automatically: lighter prompts hit faster, cheaper models; harder ones trigger “thinking” modes that burn more compute. This simplifies UX, optimizes cost/performance, and gives OpenAI more control over inference paths without forcing user-visible changes. (3) Aggressive pricing. At $1.25/million input tokens, GPT-5 matches Gemini 2.5 Pro and undercuts Claude Opus 4.1 by over 10×. Output tokens are similarly priced. This smells like the opening shots of a pricing war. (4) Tiered rollout. GPT-5 Mini is the new default for free ChatGPT users. Paid tiers unlock GPT-5 Pro and GPT-5 Thinking with higher limits and deeper reasoning. Free access drives lock-in; tiers drive ARPU. (5) Hallucinations. GPT-5 hallucinates ~9.6% with web access and ~4.5% in reasoning mode. New “safe completions” replace blanket refusals with partial, policy-compliant answers. Accuracy remains the bottleneck for full autonomy, and verification layers are still non-negotiable. Analysts call it evolutionary; vocal users call it a “downgrade” in tone and style. On Reddit, the mood is that Google will “cook OpenAI” with Gemini 3. Elon says his next model is ready. My take? The LLM arms race is converging. Whoever holds the “best model” crown by year-end will likely win by inches, not miles - and that edge will be fleeting. Calling GPT-5 incremental misses the slope of the curve: we haven’t been idle for two years since GPT-4, we’ve been compounding capabilities across the entire field. The water’s heating up fast, fellow frogs, and if you’re waiting for a single big splash, you might not notice we’re already in a rolling boil.

358

36 Comments

Elad Gil is a solo GP raising $1.5 billion. Some of his early bets include Harvey, Perplexity, Character.AI, Mistral AI, Braintrust and others. In a recent post, he says: “We’ve entered an era where the first set of AI markets have solidified.” 🔍 Which AI markets have crystallised? 1. Foundation Models (LLMs) – OpenAI, Anthropic, Mistral AI, Meta, xAI, Google Gemini. 2. Code Generation – GitHub Copilot, Cursor, Cognition, Claude Code, Replit. 3. Legal AI – Harvey, Casetext, Part of Thomson Reuters, EvenUp. 4. Medical Scribing – Abridge, Ambience Healthcare, Nuance Communications. 5. Customer Support – Decagon, Sierra, Forethought. 6. Search + IR – Perplexity, OpenAI, Google, Meta. These markets have matured from chaos to clarity, with revenue ramping fast and agentic workflows (AI doing tasks for you) reshaping how value is captured. 🧭 What’s next? Elad sees the next frontiers forming in: – Accounting – Compliance – Financial tools – Sales agents – Security – …and AI-driven rollups of legacy service businesses. He also introduces the concept of the "GPT Ladder" where new markets open as models cross capability thresholds (e.g., GPT-5 or Claude X enabling previously impossible use cases). An interesting thought is on usage based pricing; “We’re shifting from selling seats to selling units of cognition.” Read the full post in the first comment 👇

62

10 Comments

How we rebuilt every Draper investing workflow with a custom AI stack, and why any VC team needs to do the same. 👇 We tested everything: Off-the-shelf CRMs, AI deal platforms, fancy automation suites. They were rigid, pricey, and built for box-checking, not for how we actually invest: fast, via 40 years of pattern recognition, thesis-driven, intuition-heavy. So we built our own, simply using Airtable, Otter.ai, Mappa, Open AI APIs, and Zapier to no-code automate everything together. With a lean team (plus a few fearless MBA interns) we spun up an in-house AI stack that now: 1️⃣ Ingests every update email, pitch deck, and transcript of every minute of every company meeting. From the first second we meet them to portco updates and follow ons. 2️⃣ Maps each incoming company to our sector theses and flags the outliers we love 3️⃣ Summarizes founder calls, comps, and market data in one view. Shout out to Otter.ai (a Draper portco) for the integration into our workflow. 4️⃣ Scores deals against 40 years of Draper success stories. Thank you to our Mappa (a Draper portco) AI integration that codifies our unicorn, rhino, and decacorn founders into actionable heuristics. 5️⃣ Creates one source of truth for every company that touches Draper. Every company has one singular record that that auto-logs emails, notes, memos, meeting minutes, etc. topped with an AI summary that updates in real time. 🤠 Impact in 10 weeks: • Inbox triage: 4 hours → < 20 minutes • 100% real-time visibility from first touch to term sheet • Our team of interns/analysts jumping in to run deep diligence and AI builds • Our investment team noticeably spending the reclaimed hours with founders, not multiple spreadsheets ⚡️Next up: → AI that learns what we invest in, why, and surfaces the next unconventional founders → A digital twin of Tim Draper trained on his brain, his thought processes, to coach founders for what we’re looking for in investments → Custom Draper pitch-deck feedback for founders (no more mystery rejections) → Network matching across our 18K+ contacts This isn’t just a productivity tweak. It’s venture capital rebuilt from the inside out, where internal VC intelligence compounds alongside not only our portfolio, but founders across the startup ecosystem. Thank you so much to Lucas Rosillo for leading the charge in building this incredible AI system for us. Proud to have him as an AI Advisor at Draper Associates. What else should we build in our AI stack? Drop ideas below! Would love to share thoughts. #AI #VentureCapital #Startups #Automation #FutureOfVC #DraperAssociates

143

30 Comments

OpenAI launching GPT-5 is a big step forward for the company, the AI industry, and the technology industry as a whole.  The power of the model, including key features like a much lower hallucination rate as well as its strong performance on standard evaluations will rightly attract a lot of attention from the media and practitioners.  But what many are overlooking is how the GPT-5 launch strategy is a brilliant and strategic example of blitzscaling.   Blitzscaling involves prioritizing speed over efficiency to achieve lasting competitive advantage, and that is precisely what the GPT-5 launch aims to achieve for OpenAI.  (Disclosure: I used to be a board member at OpenAI, and am also an investor in the company.)   Think back to March of 2023, when OpenAI released GPT-4.  While the announcement was big news, the actual rollout was limited.  Initially, there was a waitlist for developers, and it wasn’t generally available to ChatGPT users until months later, and even then, only to paid users.   This is a classic launch strategy for a new, powerful, but expensive technology.  Make the announcement, spread awareness, but slowly ramp up usage to iron out bugs and make sure that your servers can handle the load.   This is blitzscaling in action.  OpenAI is blitzing GPT-5 into the marketplace, looking to win the market for state-of-the-art foundational models during a period in which no other company can compete.  The aggressive rollout and pricing indicate a willingness to ignore efficiency (optimizing for revenue, smoothing out the cost of ramping up compute) for the sake of winning a bigger victory.   This is only possible because OpenAI worked with its partners on this launch.  It’s not a coincidence that Sam Altman thanked OpenAI’s partners at Microsoft, Nvidia, Oracle, Google, and Coreweave, and noted, “lots and lots of GPUs working overtime.”  He also tweeted, “melting silicon has a very distinct smell”.   The GPT-5 launch also shows how blitzscaling evolves with the circumstances.  When OpenAI launched ChatGPT in November 2022, there was no real competition, which meant that blitzscaling was about making sure that as many people as possible learned about ChatGPT.  When OpenAI launched GPT-4 in March 2023, the blitzscaling goal was to demonstrate that AI could become much more powerful than people had experienced with ChatGPT powered by GPT-3.5.  With this launch, the goal is to overwhelm the market and OpenAI’s competitors and build as large a lead as possible for this generation of foundational models, which includes becoming the model of choice for as much of the entire AI ecosystem as possible.   The other good news from all of this blitzscaling is that users, customers, developers, and all of us are all getting access to an amazing product at a blazingly-fast pace.  I can’t wait to see what people do with GPT-5 to move humanity closer to superagency.

442

144 Comments

The best AI apps today are being built like infrastructure companies. Great breakdown from the awesome 💡 Yazan "Yaz" El-Baba on why technical depth, rapid iteration, and margin-aware architectures are becoming table stakes for AI startups. Definitely worth a read: https://lnkd.in/gMVFFXgH 

54

2 Comments

“Based on Nvidia's latest datacenter sales figures, AI capex may be ~2% of US GDP in 2025…as a % of GDP, spending on AI infra has already exceeded spending on telecom and internet infrastructure from the dot-com boom—and it’s still growing.” Private sector driven stimulus spending. 📈 https://lnkd.in/euvZ2qAb

87

10 Comments

This is the kind of bold bet that only founder-led companies can make: “Meta is reportedly investing nearly $15B in the data-labeling firm Scale AI and taking a 49% stake in the startup, while also bringing on CEO Alexandr Wang to help lead a new “superintelligence” lab within the company.”

46

1 Comment

Lessons from the Past – Kosmix and Innovative Search Interfaces Long before today’s generative AI revolution, Kosmix was already challenging traditional search interfaces. Founded in the mid-2000s, Kosmix aimed to organize the web around topics rather than keywords, pioneering the concept of "algorithmic topic pages." I had the pleasure of being an early team member of Kosmix. These topic pages were automated, multimedia-rich dashboards tailored to each search query, categorizing and curating information from across the web into coherent, interactive experiences. Kosmix’s categorization engine intelligently identified content types relevant to each query, creating unique, dynamic presentations. One notable innovation was "Tweetbeat," designed for real-time experiences during live events such as the FIFA World Cup. It seamlessly integrated Twitter feeds, filtering real-time posts alongside scores and updates, creating an immersive, interactive search result experience. Kosmix’s groundbreaking approach highlighted critical lessons: the value of context-aware, dynamic content presentations, the need for effective user engagement without overwhelming users, and the complexity of integrating real-time data into search results. Today, generative AI-powered search platforms leverage these insights, employing advanced AI to deliver contextually relevant, dynamic, multimodal search experiences that echo Kosmix's visionary concepts. Kosmix’s story highlights the power—and the complexity—of rethinking the search interface. In the concluding post of this series, we'll envision the future of search, reimagined with generative AI. #SearchInnovation #TechHistory #WebEvolution #UserExperienceDesign #DynamicContent #AIinSearch #InteractiveMedia #InformationRetrieval #RealTimeDataIntegration #FutureOfSearch

31

1 Comment

Three themes in the AI infrastructure and security space that founders are sleeping on: 1/ AI agent identity & security We used to think about securing people. Then machines. Now we need to think about agents. With AI agents coming online, the scope of threats is higher than ever. We need: →Agent identity management →Guardrails for autonomous actions →Security frameworks for agent-to-agent interactions There's been a lot of talk about general AI security, but agent security is a completely different beast with much higher stakes. 2/ Data collection from end users Everyone obsesses over computers. The real moat is data. Higher quality data = better models = better companies. But data now includes new forms beyond traditional training sets: →RLHF data from product usage →RL data from steps and actions experts take in their work If you can build systems that collect unique datasets through actual product usage, you're building a defensible business. 3/ Feedback loops that expand your user base Most infrastructure tools are built for the 5-20 AI engineers on a team. But if you want big contracts, you need to sell to managers and execs. Create interfaces for the broader organization like: →Dashboards for analytics →Approval workflows →Compliance visibility The magic happens when your core technical product has feedback loops that make non-technical people want to use it too.

111

28 Comments

7 steps to modeling the Pro Forma Balance Sheet (+ template) 👇 ~~~ 📌 𝗧𝗟;𝗗𝗥: grab the template here 👉 https://lnkd.in/ex6BKa4E ~~~ 𝟭. 𝗦𝗲𝗹𝗹𝗲𝗿 𝗧𝗮𝗸𝗲𝘀 𝗖𝗮𝘀𝗵 𝗮𝘁 𝗖𝗹𝗼𝘀𝗲 Seller keeps the existing cash because it's value they previously created and doesn't belong to the new buyer. (Often times this is 𝘦𝘹𝘤𝘦𝘴𝘴 cash, not all cash) Steps: Reduces cash, reduces Retained Earnings. 𝟮. 𝗡𝗲𝘄 𝗘𝗾𝘂𝗶𝘁𝘆 Buyer brings new equity to buy the business. Steps: Equity up, Cash up 𝟯. 𝗡𝗲𝘄 𝗗𝗲𝗯𝘁 Buyer also brings new debt to purchase the business. Steps: Debt up, Cash up 𝟰. 𝗣𝗮𝘆𝗼𝗳𝗳 𝗢𝗹𝗱 𝗘𝗾𝘂𝗶𝘁𝘆 (𝗮𝗸𝗮 𝗯𝘂𝘆 𝘁𝗵𝗲 𝗯𝘂𝘀𝗶𝗻𝗲𝘀𝘀) The cash from the new equity & debt goes to buy the existing business. Any premium above the equity value is recorded as Goodwill. Steps: Old Equity Paid Off, Goodwill Calculated, Cash goes out (this is "The Purchase") How to calculate Goodwill: Enterprise Value ( + ) Cash ( - ) Debt = FMV of Equity ( - ) Book Value of Existing Equity = Goodwill 𝟱. 𝗣𝗮𝘆𝗼𝗳𝗳 𝗢𝗹𝗱 𝗗𝗲𝗯𝘁 Similar to step 4, new debt comes in, old debt goes away. Steps: Reduce old debt, reduce cash 𝟲. 𝗧𝗿𝗮𝗻𝘀𝗮𝗰𝘁𝗶𝗼𝗻 𝗙𝗲𝗲𝘀 These are an expense for the new buyer (to get the deal done), so it immediately hits Retained Earnings. Steps: Retained Earnings down, Cash down 𝟳. 𝗗𝗲𝗯𝘁 𝗙𝗲𝗲𝘀 Under the "old method," you would record the deferred financing fees as an asset. Today they're recorded as either a contra-liability or asset depending on the type of debt. The impact on the P&L is the same (amortization), so I often prefer the "old method" to keep my model cleaner. Steps: asset up, cash down 𝗪𝗮𝘁𝗰𝗵 𝘁𝗵𝗲 𝗖𝗮𝘀𝗵 In every step, you can see how the cash immediately flows in and subsequently flows out in order to complete the purchase. The ending result is the "Pro Forma Balance Sheet" that reflects the new debt, new equity, and cash balance for the buyer ($500k in this case). ~~~ 👋 Hey! I'm Chris Reilly. 🟢 I can help you become the go-to Financial Modeler in just 8 hours. ⏩ https://bit.ly/FMECourses

85

15 Comments

𝗔𝗜 𝗔𝗴𝗲𝗻𝘁𝘀 𝗔𝗿𝗲 𝘁𝗵𝗲 𝗥𝗮𝗴𝗲, 𝗦𝗼 𝗪𝗵𝘆 𝗔𝗺 𝗜 𝗟𝗼𝗼𝗸𝗶𝗻𝗴 𝗳𝗼𝗿 𝘁𝗵𝗲 𝗡𝗲𝘅𝘁 𝗖𝗥𝗠? AI agents are everywhere. They promise to schedule meetings, write follow-up emails, research leads, and even close deals on your behalf. The vision is compelling: a future where software acts autonomously, taking care of work while you focus on higher-level decisions. But here’s the thing—I’m not chasing the next agent. I’m hunting for the next generation of software. That might sound unsexy compared to the sweeping autonomy of agentic AI, but if history is any guide, it’s not agents that reshape industries—it’s applications. Tools like CRMs, email, spreadsheets, and calendars are the real engines of work. And every time computing undergoes a paradigm shift, it’s these tools that get rebuilt—and in doing so, they redefine how we work. When the web overtook client-server, Siebel patched its software while Salesforce reimagined it. When mobile became dominant, the winners weren’t the ones who augmented with an app, but those who rethought entire products—WhatsApp, not Skype; Instagram, not Flickr. Now with AI, we’re seeing the same dynamic: incumbents layering generative features on top of legacy systems—“AI-powered” summaries, data-entry assistants, and chat bots—but all within rigid paradigms from a pre-AI world. Even as tech giants like Salesforce and Google pour billions into AI and acquire early-stage disruptors, they still can’t rewrite the core assumptions baked into their legacy products. It’s lipstick on a pig. Yes, they have distribution—but in every platform shift, distribution advantages erode when the foundation of the product changes. The true innovation comes not from bolted-on features, but from rethinking what the product is. The opportunity isn’t adding AI to CRMs. It’s entirely reconceptualizing how all software works—starting with the tools we rely on every day. Imagine a CRM that doesn’t just log data but generates it—hunts for your leads, drafts your outreach, bids with context, preps you for meetings, and coaches you in real time. A spreadsheet that interprets intent and self-builds models to expose insights to you. A calendar that understands not just when you're free or busy, but that a regular 1:1 can be moved to make room for a board meeting—or that a visiting exec with limited time should take priority over a standing internal check-in. That’s what AI-native looks like—not just automation, but a fundamentally new interaction model that reshapes the flow of work and decision-making. AI agents will play a role in this future, but they’re not the destination—they’re ingredients. The magic happens when they’re orchestrated into cohesive product experiences built for how we’ll actually work in an AI-native world. So yes, agents are impressive. But I’m looking beyond the next agent. I’m looking for the next generation of applications. Monk's Hill Ventures Phong T. Bea Ramos

56

1 Comment

Generative AI has the potential to completely remake the advertising and media ecosystem. The question is: To what end? What happens in a "post traffic" world? This was the topic discussed last week Human Ventures HQ (aka Casa Komos). Algorithms (AI) in the hands of the major platforms have shaped our information ecosystem and the attention economy for decades. And, in turn, have shaped society for better and, too often, for worse. This moment is either a chance to reset the role algorithms play in people's relationship to media brands or completely break media as we know it. Here are just a few of many insights shared (there are a lot more, but I hit the LinkedIn character limit. Will put more in the comments): “The keyword industrial complex is dead. Now it’s about intent with context—AI doesn’t think in keywords, it thinks in intent. How can you figure out the different need states of consumers as an agency or as someone working with a brand? How do you build content for that? Optimizing these channels and building brand agents are the next thing. The future will be about optimizing content and answers for the LLMS and brand agents to interface. It’s no longer about winning Share of Voice. We are entering a world where a brand needs to win Share of Model.” Joel Lunenfeld, Publicis “A lot of people are blaming AI for weaknesses that have existed in business for a long time—the audience has come last because you had to write for algorithms and then because of the way of the system you had to have a terrible user experience. So it’s going to be publishers treating themselves as brands and getting back to brand marketing. And that part I think will do quite well.” —Brian Morrissey, The Rebooting ”We’re all at the same starting line. If anyone feels like they are ahead or that they have it all figured out, I think you’re BS-ing us. Some people are going to be sprinters, some people are going to be long distance, some people are going to come from the outside, some people are going to fall—some people are going to get to the finish line and some people won't. We’re at a time and an inflection point where we need to share and learn from one another.“ —Sherry Phillips, Forbes “The whole customer discovery process has changed. As a marketer and an advertiser, we need to figure out how to get in the middle of the discovery process. That’s our job. to become part of that process. And if we’re not talented enough to figure that out, then we shouldn’t be in the business we’re in. So how do you actually create content? How do you actually create interactions? How do you create discovery engines? How do you create agents that actually bring value and trust between you and your consumer? That's ultimately where you need to go.” —Bob Lord, Horizon Media “If keywords are dead, does that mean block lists are dead too? Because that would be great for us. And those words are crazy because there's how many of them now right?" —Jessica Sibley, Time

248

33 Comments

This startup just raised 187 million dollars to make medicine in space 🤯 The company is called Varda. They’ve now raised 329 million dollars from top investors like Peter Thiel, Khosla Ventures, and Lux Capital Here’s what they’re doing: 1. They’re building small factories that work in space In zero gravity, you can make drugs that are purer and more effective than the ones made on Earth That’s what Varda is focused on 2. They’re already sending missions to orbit Their next launch is planned before the end of this year Each mission brings the products back down to Earth with their own custom capsules 3. They want to build a real business out of this Not just experiments Real products, made in space and sold on Earth 4. Some of the world’s smartest investors are backing them These are people who have built and funded companies like SpaceX, Palantir, and OpenAI They think Varda has a shot 2025 is getting pretty wild 🤯

86

19 Comments