Andreessen Horowitz

Interactive Brokers accounts trade far more than other retail accounts, with ~4x the daily trading volume.

Marc Andreessen’s career advice to Americans in their early twenties:

AI continues to be the everything-story for everyone. “IT Capex” is approaching 40% of all Capex for the entire S&P:

We partnered with Pylon to investigate whether AI is replacing or augmenting jobs in customer support. The resounding answer: AI is acting much more as an invisible triage agent or copilot than it is as an end-to-end support handler. If your company as a whole is pretty AI-native, there’s a much higher likelihood that your customer support AI will be good at resolving questions end-to-end. In particular, “AI-resolved” tickets are more likely to be actual resolutions, as opposed to “the customer gave up and dropped off.” But if you’re a really important, big-spending B2B customer? You’re probably still gonna get a human. Full breakdown linked in the comments.

From the beginning at Pylon, we've had this conviction that you can't just run B2B support on platforms made for B2C. Why? Account context. Context is everything when your customers... → Are accounts with multiple stakeholders → Sign multi-year contracts → Pay you 6+ figures in ARR → Form long-term relationships with your team → Talk to you via unstructured channels like Slack Given the above it's impossible to reduce B2B support interactions to one-time transactions and you can't afford to "forget" customer context and history. This is why we built Account Intelligence. We're the first support platform to connect support data back to account-level context, because we've seen that building up your customer knowledge graph consistently improves outcomes. We partnered with the team at Andreessen Horowitz to prove it with data directly from Pylon. When you maintain and leverage rich context on your accounts, you improve AI-driven resolution and your customers are more satisfied with the support they're getting. We did a whole deep dive on Pylon data and wrote about why contrary to the popular narrative, AI is best as a copilot for support teams and not a replacement for them. Link to the full piece in comments 👇️ — Huge thanks to Jared Forman, our Head of Data, and Jason Saltzman + team at a16z for the analysis.

Introducing the a16z FDE Fellowship: An 8-week cohort for FDEs and Applied AI leaders at the forefront of deploying AI into real-world enterprises. Demand for FDEs is growing rapidly as startups race to close the gap between what customers need and what AI agents can deliver. But best-in-class FDEs are rare and hard to reach. We’re launching this fellowship to bring together the people at the center of this shift to share what’s working, level up, and define the future of forward-deployed engineering. If this sounds like you, apply here by Friday, June 26th, 2026. Link in the comments.

Marc Andreessen on venture capital's 2x2 matrix: "You make all your money on successful and non-consensus... if something is already consensus then money will have already flooded in and the profit opportunity is gone." His translation of non-consensus: "in sort of practical terms, it translates to crazy. You are investing in things that look like they are just nuts."

When Marc Andreessen and Ben Horowitz founded Andreessen Horowitz, or a16z in 2009, it was built on the idea that software was eating the world, shaped by their experiences in the industry. And coders have a thing for naming shortcuts:

Join us for the inaugural Chief of Staff Bootcamp, this June 22nd in San Francisco. This is a closed-door, one-day working session for a small, curated group of Chiefs of Staff across a16z portfolio companies and our broader community. This program enables Chiefs of Staff to share how their companies actually run planning, GTM, product, and operating cadence and return with concrete documents and insights to improve execution immediately. Attendees can expect candid conversations, speaker presentations, breakout sessions, and peer exchange, along with concrete documents and insights to improve execution immediately. Application is linked in the comments.

OpenAI and Anthropic are effectively telling the market they can't solve every problem with a generic AI coworker. You don't pour billions into massive forward-deployed joint ventures if you think the next model release is going to take care of it. In the cloud supercycle, semis led and software followed (and you didn't need Qualcomm or ARM to tell you the value was migrating up the stack). In AI, the infra layer itself is telling us the application layer is a separate, massive opportunity they can't fully capture. Full piece by Joe Schmidt IV linked in the comments.