Under the impact of OpenClaw, Cursor is already becoming outdated.

Source: Founder Park

“Cursor is F***ing and Everyone Knows It.”

Insight Partners Co-Founder Jerry Murdock told 20VC this.

Founded in 1995, Insight Partners is a leading global VC/PE firm focused on software and internet sectors, managing over $90 billion in assets. Its portfolio has shaped the modern software economy. Jerry Murdock rarely participates in podcasts, making this his first in-depth interview.

The wave sparked by OpenClaw has, in just two months, rapidly impacted and diluted the value of SaaS giants. Not only SaaS, but even Cursor is becoming outdated.

In this conversation with 20VC, Jerry shared many seemingly “non-consensus” views:

Autonomous Agents are the core of this tsunami, not just broad AI.

Cursor is outdated; it only represents the early stage.

We will inevitably see a dedicated tech stack for Autonomous Agents. Once the orchestration layer is solidified, Agents will directly determine computing resource flow. In tech selection, autonomous Agents will hold absolute influence.

Currently, all software is ultimately purchased by humans, but in the future, software will be bought and used by Agents. Whatever your business, ask yourself: will this be sold to humans or to intelligent entities?

Agents will make record-keeping systems more important, or worthless.

Perhaps, under the impact of OpenClaw, all AI companies will need to return to the same “basket” and restart.

Similarly, Cursor’s CEO Michael Truell recently posted a long tweet about their transition toward “autonomous Agents”:

We are now entering the third era of AI software development, where agents can independently handle larger, more complex tasks, work longer hours, and require less frequent human guidance.

Therefore, Cursor’s core function is no longer just “coding,” but helping developers build a “factory” that can produce their software.

Below are the highlights of the Jerry Murdock and 20VC discussion.

1. Autonomous Agents are the core of this tsunami, not just broad AI

Host: Many companies think Cursor is outdated. You described this wave of AI as a “tsunami”—what does that mean specifically?

Jerry Murdock: If I have any experience, it’s from repeatedly messing up and learning lessons. A tsunami in deep ocean isn’t threatening; the real danger is when it hits the beach, and it’s chaotic.

What’s coming isn’t just a single product but autonomous Agents. That’s the core of this tsunami, not just broad AI.

Host: What stage are we at now? Is this the “end of SaaS” everyone talks about?

Jerry Murdock: I’m not a doomsayer, but change is coming too fast, and we must take the initiative. Many companies think, “I just need to add AI to my existing business,” which might help them survive and cash out, but only those with “AI Native” thinking will become truly outstanding. The open-source communities emerging en masse are the real force of impact.

2. Cursor is outdated

Host: In your portfolio, what changes are still unnoticed by the outside world?

Jerry Murdock: If I look at real AI startups like E2B, Eventual, Lotus AI, they’re already using OpenClaw or self-developed autonomous intelligent agents to write code in practice. And this has happened in less than two months. That’s the incredible part.

Host: Yes, they’ve started letting agents write code themselves. What does this mean for a company like Cursor, valued at two or three hundred billion?

Jerry Murdock: The most cutting-edge companies I mentioned believe Cursor is outdated; it’s just an early-stage product now.

Of course, the Cursor team is smart, well-funded, and has many clients. They have time to embrace autonomous intelligence and pivot to the next direction. But in AI, you must move toward future trends, not stay in the past.

3. We will inevitably see a “Claw tech stack”

Host: What impact do you think OpenClaw and the open-source community will have on a larger scale?

Jerry Murdock: Today’s open-source community has many people doing integrations. If this community accelerates, we’ll see agents achieve capabilities that are currently impossible.

We will inevitably see a “Claw tech stack,” or a dedicated tech stack for autonomous intelligent agents. Just like the LAMP stack (Linux, Apache, MySQL, PHP) in 2004 drastically lowered website development costs and sparked a boom in websites and e-commerce.

Now, Claude, Codex, and Gemini dominate reasoning layers. But eventually, there will be an orchestration layer where agents can schedule multiple different LLMs for workflows: for complex tasks, call Claude; for simpler parts, distribute to open-source models like DeepSeek or Llama 3.

Once the orchestration layer is stable enough, agents will directly decide the flow of computing resources. I believe this will effectively boost the rise of open-source models and increase demand for ASIC chips, as ASICs can embed models directly on chips, reducing costs and being more suited for specific workloads than general-purpose GPUs. Overall, this is a revolution.

Host: If computing power shifts to ASICs, will Nvidia’s value shrink?

Jerry Murdock: It depends on execution. Jensen Huang’s acquisition of Groq is to gain the ability to embed memory directly on chips, ensuring future CUDA supports ASICs. Many criticize Meta for falling behind, but Zuckerberg daring to tell Huang, “Sorry, we don’t need you,” shows he’s heavily betting on ASICs.

Host: Will dynamic allocation among models fully commoditize large models?

Jerry Murdock: Ultimately, the decision-making will be in autonomous agents, not developers. Developers will choose chips based on experience and ideas, but agents are probabilistic. They will directly call multiple Python libraries, run in different sandboxes, and pick the best performance. So, in the future, autonomous agents will hold more influence.

4. Agents will make record systems more important or worthless

Host: Will the explosion of autonomous agents make current record-keeping systems worthless or more valuable?

Jerry Murdock: It depends. Take Carta, for example. If the tokenization of stocks becomes mainstream and the market uses Carta’s system for management, Carta’s value will skyrocket. But if everyone bypasses Carta with agents and builds new systems, its future is uncertain.

Salesforce is like Everest in enterprise software; it won’t melt overnight. But the perspective on its valuation has changed: you need to look at the thousands of companies built on Salesforce. If they start collapsing, Salesforce’s underlying value will shrink too.

Host: Revenue, growth rate, profit margin… When these familiar valuation metrics become unreliable, where is a company’s value?

Jerry Murdock: It depends on management’s ability to adapt. Most software companies tend to expand first and contract later when new tech emerges because people are used to inertia and doing what they’ve always done. But I don’t think ordinary software companies will disappear.

If you own a record system company with vast data and can leverage autonomous agents to utilize that data, its value will soar. Conversely, if you react slowly and don’t understand the new market, software will decline. The tsunami is a warning—when it hits the beach, don’t stay on the shore; run to higher ground.

Host: With OpenAI, Anthropic, and Google, which would you choose as an ambassador?

Jerry Murdock: It depends on the stage of opportunity you give me.

Host: I offer you $50 billion valuation OpenAI or $38 billion valuation Anthropic.

Jerry Murdock: OpenAI. Because it already has 800 million users.

Crossing the billion-user mark is a success. Google’s products also have a billion users, but if I can turn OpenAI’s chatbot into my own personalized “Jerry Murdock AI,” that’s an extremely great business.

5. Software buyers are shifting from humans to agents

Host: Is the high-growth SaaS model we’re used to outdated?

Jerry Murdock: Think about it—today, all software is ultimately bought by humans, but in the future, it will be bought and used by agents.

An autonomous agent will become your “employee.” You give it credentials and permissions, and it makes decisions. You’ll review it like managing employees: “What did you buy? How much did you spend? What did you do?” So, whatever you invest in, consider: will this business be controlled and used by intelligent entities in the future?

Host: What happens to business models when buyers shift from humans to agents?

Jerry Murdock: This has never happened before. I believe “consumption-based” pricing will become mainstream. For example, Docker shifted toward AI by adopting this model. If agents are authorized to use sandboxes, you pay based on actual consumption. When limits are nearly reached, agents will proactively seek your approval for the next step. That’s the future operation mode.

Host: Many SaaS companies are trying “bolt-on” AI strategies for transformation. Do you think it works?

Jerry Murdock: Like competing in the Olympics—everyone wants gold, but only a few win. You can try adding AI as a bolt-on, but to win, you must achieve AI Native, world-class standards.

Autonomous agents write software faster and cheaper than humans, and they’re already doing it. So, if you develop software for autonomous agents, and they have reason to use it, and it’s valuable, that’s great—you’ll do well. But if you haven’t started now, you’ll face challenges in 6 months, a year, or 18 months. If you still think humans will buy your software then, you’re in trouble.

6. Autonomous Agents are evolving into true AI employees

Host: If in the future, agents buy software and make decisions, what impact will that have on the labor market?

Jerry Murdock: This will be a core issue in the next few years, even affecting the next election. White-collar jobs like administrative assistants, customer service, junior marketing—most will be better done by autonomous agents.

Companies will stop or slow hiring for these roles. The first to be threatened aren’t current employees but the entry-level developers, admins, or customer service reps they planned to hire.

They’ve already demonstrated this capability: writing code, handling scheduling needs. They can replace humans and will do so.

I believe the job market is highly uneven. The first adopters will be small and medium-sized enterprises, where a secretary can make a huge difference. In companies with just a few people, a secretary can significantly change how they handle customer service.

We should watch the growth of autonomous agents in three areas: consumers, small businesses, and large enterprises. I think large companies will be last. They’ve been slow in this AI revolution.

Also, “Universal Basic Income (UBI)” could become a reality in two and a half years or at least a key voting issue, as governments must provide safety nets for unemployed populations.

Host: Klarna’s CEO said they had 7,000 employees at peak, but by 2030, it will be fewer than 2,000. Has a large workforce become a kind of “liability”?

Murdock: It’s about culture. Steve Jobs talked about this at Apple: how to keep only A-grade talent. In the future, your company will have far more intelligent agents than humans. The question is, what are the remaining 2,000 doing? What’s their culture? If they gain a higher quality of life and more time with their families, that could be a huge victory.

Host: Do you think a single-person company with a billion-dollar revenue will appear?

Murdock: Absolutely. It depends on how smart your agents are, how well you deploy them, and how willing you are to listen to them.

What’s truly shocking is that autonomous intelligent agents are effective—they don’t need constant oversight and can work independently. We’ve evolved from having an “AI assistant” to having a “real employee.” This is a fundamental change in business operation models.

Host: Given the tough environment, what specific advice would you give to recent graduates entering the workforce or looking for their first job?

Jerry Murdock: I’d tell them to buy a Mac Mini, install OpenClaw, and bring their OpenClaw to interviews.

Host: Bring an agent to the interview?

Jerry Murdock: Yes. The amazing thing is, autonomous intelligent agents are truly effective—they work independently without constant supervision. We’ve moved from having an “assistant” to having a “real employee.” You need to demonstrate not just your personal skills but your ability to use and amplify with agents. That will be your competitive advantage in the future.

7. Now is the best time to start a new fund

Host: If I’m a traditional private equity investor holding SaaS companies growing 15-20% annually, what should I do?

Jerry Murdock: After 9/11, top acquisition firms like Forstmann Little bet heavily on old models and eventually exited completely. Some institutions will follow suit. But now is the time to revisit past assumptions and future decisions.

Host: In today’s tech landscape, if you were to start an investment firm from scratch, how would you do it?

Jerry Murdock: Undoubtedly, future VC and PE firms must have their own autonomous agents—everyone needs one.

If I started from zero, the ideal would be having excellent data to use agents to identify market gaps that new AI companies can fill. Next, the evaluation should focus not only on entrepreneurs but also on their ability and quality of using autonomous agents. VCs and entrepreneurs are on the same starting line, competing to see who can better harness these agents.

Host: Is now the best time to start a new fund?

Jerry Murdock: Absolutely. We are in a period of great change—humans will no longer be the decision-makers in software; autonomous agents will be. This tsunami is unlike anything before. Those who embrace the new model early can start from zero and have a huge advantage over those successful in the old ways but slow to adapt, because they are already wealthy and have made a lot of money.