Andreessen Horowitz

Sat down with Isa Fulford and Christina Kim to break down the GPT-5 Launch. Below are the 15 main differences between this model and the old: 1. Smarter Reasoning: GPT-5 shows major improvements in reasoning, instruction following, and multi-step thinking. 2. Best-in-Class Coding: It is now the top coding model, especially strong in front-end web development. 3. Creative Writing Boost: Writing is more human, emotional, and stylistically nuanced for both casual and serious tasks. 4. Healthier Assistant Behavior: Less sycophantic, more thoughtful, and better balanced in tone and helpfulness. 5. Agent Integration: Powers agents that can browse, take actions, and generate work artifacts like documents, slides, and spreadsheets. 6. Asynchronous Execution: Handles background tasks over time such as research reports or planning, not just instant answers. 7. Mid-Training Phase: Adds a new training step between pre-training and post-training to expand knowledge and skills. 8. Efficient Learning: Learns new capabilities quickly from small, high-quality datasets using reinforcement learning. 9. Better Front-End UX Output: Generates cleaner and more usable front-end components for web interfaces. 10. General-Purpose Utility: Designed to be useful for a wide range of users including developers, analysts, and everyday people. 11. Improved Browsing: Performs deeper and more reliable web research using enhanced browsing capabilities. 12. Tool Use Across Modalities: Can interact with terminals, browsers, and visual interfaces such as screenshots. 13. Higher User Expectations: Users now tolerate longer wait times in exchange for more valuable results. 14. Customized Internal Evals: OpenAI uses task-specific evaluations like spreadsheet editing to guide development. 15. Emphasis on Data Quality: Prioritizes carefully curated datasets over brute-force scaling or architecture changes.

The Benefits of Operating with an Owner Mentality The biggest growth accelerator in my career hasn’t been a promotion or a big project — it’s been choosing to think like an owner, no matter the size of the company. Most people assume an owner mentality matters more in a startup than in a big company. I’ve found it’s exactly the opposite. In large companies, so few people operate this way that if you do, you stand out faster than you can imagine. I believe it can be a catalytic property for your career. Here are 7 big reasons why an owner mindset changes everything about the way you operate: ⸻ 1. Your tenacity and tempo instantly change. When you think of yourself as an owner versus an employee, you start to think from the market inwards and get deeply obsessed with your customers. You move faster because every owner believes they can move faster. You’re always slightly dissatisfied, never accepting mediocrity in yourself or your work. ⸻ 2. You stop building fiefdoms and start building momentum. Owners don’t cling to scope or titles; they focus on what will make the company win. If giving up a function, a team, or even a cherished project puts the company in a better place, you do it — without hesitation. ⸻ 3. You care about the whole business, not just your lane. An owner doesn’t only optimize for their department’s metrics. They think about the entire system — customers, revenue, brand, culture. If something is broken anywhere, it’s your problem. ⸻ 4. You stand out faster in a large company than you ever could in a small one. Here’s the counterintuitive part: in a big company, so few people operate this way that if you do, you’ll get noticed almost immediately. High ownership is rare — which makes it incredibly visible. ⸻ 5. You’re secure enough to own failure and share credit. When something goes wrong, you take the blame. When something goes right, you give the credit away. People trust owners who aren’t insecure about recognition — and that trust compounds over time. ⸻ 6. You attract other high-ownership people. This mindset is contagious. When you take accountability and push for impact, others around you start doing the same. Over time, the whole culture shifts — faster, bolder, hungrier. ⸻ 7. Your personal growth is more fulfilling ‘cause you are building something bigger than yourself. When you operate like an owner, you naturally take on bigger challenges than your role requires. You learn faster, build broader skills, and expand your thinking. ⸻ It’s tempting in a big company to play it safe, stay in your lane. But the leaders who grow fastest — and have the most impact — are the ones who choose to think and act like owners, even without the title. In the long run, the company benefits. But you benefit even more. This was just one of the topics I got to explore in a conversation with two of my good friends, Martin Casado and Raghu Raghuram. Full link of podcast in comments.

Permissioned data access appears to be under siege. If data access becomes a closed-loop system, what breaks — and what is lost? I’ve met hundreds of B2B AI startups. Here's how I think this reshapes the landscape and how founders can respond 👇 https://xmrrwallet.com/cmx.plnkd.in/e_xAZZ9p

AI Agents Are No Longer Experimental. They're Becoming Core Strategy. In Q4’24, enterprises were testing AI. By Q1’25, they’re reorganizing around it. We analyzed earnings calls from 19 leading public companies, spanning hyperscalers, infrastructure companies, and app-layer giants, to understand how they're evolving to maintain a competitive edge. Here’s what stood out: - Agentic AI is showing up across product roadmaps. Companies are embedding AI directly into core platforms and increasingly are referencing autonomous agents in their go-forward plans. - Usage-based pricing is gaining ground. Many are offering AI credits within subscriptions and monetizing increased usage through credit packs or pay-as-you-go. - Advances in reasoning models like o3 and Grok4 are setting the stage for agents that can handle complex, multi-step tasks. - Standalone AI SKUs often reflect deeper agentic functionality and are being priced accordingly. - Unified customer data is emerging as a key differentiator, and some incumbents are restricting third-party access as competition intensifies. - M&A is being used to accelerate AI capability-building. Recent acquisitions like Moveworks and Windsurf reflect interest in acquiring AI-native teams. - Infra vendors are launching new products built specifically for AI in deployment, not just embedding AI into existing tools. - Agentic systems are creating new monitoring and security demands, leading to fresh product lines in LLM scanning and observability. - Multi-model access is now a core selling point, with companies highlighting support for both proprietary and third-party models. - Hyperscalers are aiming to differentiate with silicon and tooling, from TPUs and Trainium2 to open-source agent development kits. AI is becoming a central force reshaping product strategy, infrastructure investment, and enterprise GTM.

Marc Andreessen in 1996 with Charlie Rose: "The internet is everywhere. In 10 years, we probably won’t even be talking about the internet." Everyone connected. Any network. Infinite use cases. The internet became the substrate. Everything else followed.

Startups don’t win by defying the trend. They win by riding it early and building with clarity. " Make sure that everything that you do actually has AI as a tailwind. Otherwise, you’re not going to win." Timing beats everything. Market beats team. Product only matters if it reaches scale. This is Jeetu Patel's framework for building great companies.

We’re leading Rillet’s $70M Series B. Accounting is a $500B market still stuck in the 1990s. Most ERP systems today were designed before real-time data, APIs, or AI copilots existed. Finance teams are forced to work with legacy tools that slow them down. Rillet is taking a fundamentally different approach. They’re building the AI-native ERP finance teams actually want to use. It’s fast, structured, and live in weeks (not months). Customers are closing books in days, automating manual work, and getting real-time visibility into the numbers that drive the business. And we believe the timing has never been better to reinvent this space: - AI is finally enterprise-grade - CFOs are being asked to do more with less - New infrastructure makes real-time finance possible Congrats to Nicolas Kopp and the entire Rillet team.

Marc Andreessen says embodied AI is already here. From autonomous drones to self-driving cars, and it's just the beginning. “ The general purpose robotics thing is going to happen, and it's going to happen in the next decade… at giant scale.” The U.S. has a choice: lead the Fourth Industrial Revolution, or live in a world of Chinese robots.

Everything technology touches collapses in price. Marc Andreessen says that’s why TVs, video games, basically anything with electronics in them, keep getting cheaper, while housing, healthcare, and education keep rising in price. “Technology hasn’t really touched them… Productivity growth hasn’t happened in them.” The challenge now: bring technology where it’s needed most.